Unlike larger WAN optimization solutions that you might use in data centers and ROBOs, CacheFront should work no matter where the end-user is connecting from. It does need to be installed on endpoint devices, but only on endpoint devices. What gets cached is user selectable, the caches can grow as large as needed, and include some good things like built-in security and remote "wipe" features. Today CacheFront is limited to Windows IE browsers, but interoperability should broaden out in the next year.

CacheFront Professional (also recently announced) and the orginal solution Enterprise also contain a synchronization engine. This goes beyond passive caching to actively synchronizing designated web pages (sites, documents, et.al.). For some highly mobile and often disconnected users like sales folk that want to be able to get to collateral libraries or pricing lists on the road or in the air, this could be magic. 

CacheFront Enterprise goes all the way to enabling local editing of docs and web forms (and can support custom web apps with some configuration), enabling many browser apps to run as if actually live connected (queued for updating when reconnected). This is pretty sweet, and we imagine SharePoint, Salesforce and other SaaS users will be demanding this once they become aware it is possible.

More than a cache, this really a way of virtualizing browser applications, and we think as more applications become web-based, that this kind of optimization and acceleration technology will become ubiquitous. 

What is the solution? Enterprise file collaboration (EFC): cloud file collaboration with 1) high scalability, 2) security, 3) IT control, 4) usability for IT and users, and 5) compliance.

Taneja Group is publishing a report explaining market challenges, the Five Factors, and how leading vendors fit into this market landscape.

To reserve your free copy of this 15-page report, email author Christine Taylor at christine.taylor@tanejagroup.com.

As one of a small handful of certified VMware vCloud Datacenter providers, Bluelock has been an innovator in the VMware space for a number of years. For example, Bluelock worked with VMware to develop the original vCloud Connector offering, which among other things, allows customers to readily move workloads between private and public clouds. With this new RaaS offering, Bluelock demonstrates that the time for cloud-based DR services has come.

Why are customers increasingly looking toward DR solutions in the cloud? For one, the cost of downtime is higher than ever: disruptions to a company’s operations will result in lost productivity, and when those operations are customer-facing, downtime will likely lead to lost business as well. Unfortunately, existing DR solutions tend to be complex to manage and expensive, which make them less attractive to many enterprises. This has led customers to demand more reliable and cost-effective DR solutions. Given its increasing adoption and level of maturity for enterprise workloads, the cloud offers an ideal platform for multi-site (geo-distributed) DR solutions.

A number of early cloud-based DR solutions have focused on individual VM-level recovery, but have not enabled rapid and cost-effective application recovery. Bluelock RaaS, on the other hand, is focused on application recovery, ensuring that “consistency groups” of related VMs are locked together and recovered from the same point in time. This allows all resources to come back in sync without losing transactions. Unlike traditional replication approaches, the DR solution is storage agnostic and employs best-of-breed virtual replication technology, sourced via a partnership with Zerto. This enables RaaS to be non-disruptive: replication occurs at the hypervisor level, so no agents or application re-booting are required.

Bluelock provides both an “in cloud” option for existing customers (in which recovery happens in the cloud) and a “to cloud” approach for new customers that want to recover workloads hosted in their own facilities. RaaS has been in beta for a number of months, and a number of Bluelock customers have already adopted the solution. By crediting customer accounts for resources used during testing, Bluelock encourages customers to test RaaS twice per year, which in turn will increase customers’ confidence in the solution. Bluelock also allows customers to use VMware Cloud Credits to redeem RaaS.

We think Bluelock RaaS will compare quite favorably to existing DR approaches. Compared to traditional “warm site” approaches, for instance, RaaS will deliver similar RTOs and RPOs but for half the cost. And versus “cold site” solutions, we expect Bluelock RaaS will offer vastly superior RTOs and RPOs for a similar cost.

Based on what we’ve seen, Bluelock RaaS sets the standard as one of the most reliable and cost-effective recovery offerings available from service providers today. Given market forces, we believe other providers will try to emulate this offering, especially for other (non-VMware) virtualization platforms. In the meantime, Bluelock RaaS will give VMware customers one more reason to consider Bluelock for their enterprise-level cloud services needs.

"Software Defined" Defined
In particular, just because some capability is written in software doesn't make it software defined. In my thinking, software defined networking (SDN) laid the truest foundation for the term. In SDN there is a clear separation of the "data plane", where the actual data flows and functionality are delivered, and the "control plane" where the configuration and management can be dictated. There is still highly optimized, specifically featured, and widely varying hardware handling the data flows, but control of the whole "system" can be programmed in remotely through API's. Once you've made the infrastructure programmable in this way, you can work on adding all sorts of operational intelligence and dynamic behaviors with layered solutions. The implication is that you layer on this intelligent management through remote "software" control.

To be clear, the data plane functionality could be delivered by software implemented solutions. We'd expect that software implemented infrastructure like virtual machine based storage "appliances" can more readily be written to be "software definable" (being written in software, we imagine that developers would have to be cruelly restricted to not enable programmable interfaces right from the start), which might lend to some of the confusion.  But not all software implemented technology is software definable/programmable.  

EMC ViPR
With only scant early information, it seemed as if ViPR might have been just another attempt at making InVista work out. While that attempt at virtualization was aimed at competing with IBM SVC et.al., it presented a lot of challenge and conflict within EMC by essentially devaluing the primary EMC infrastructure portfolio. But with this more detailed announcement, it's clear that EMC has taken the time to develop a value-building SDDC offering, announced today as ViPR. While we await a far more in-depth look and hands-on validation of ViPR (it rolls out second half this year into GA), today's public announcement fits better into our interpretation of what "software defined storage" means and how it can add value to the future data center.

ViPR is intended to layer on top of the wide range of the whole EMC storage product portfolio and both 3rd party and commodity storage (assuming they have programmable API's). It has two identifiable components

1. ViPR Controller that layers directly over underlying storage with unified management controls like cloudy elastic provisioning, self-service storage catalog, and embedded automation. The controller layer essentially pools storage assets and unifies management (goodbye array specific SRM!), while still exposing and leveraging the specific features of the storage arrays in the data plane.
2. ViPR Data Services which will be able to "inject" certain capabilities into the data plane. 

Does the Data Services Layer simply become a new "place" to host enterprise storage features (software implemented storage capabilities) - in other words, reducing the underlying enterprise storage investment into zombie JBODs? This is definitely not EMC's intention this time around. ViPR is aimed to preserve storage hardware and software investments, and extend the usefulness of existing storage arrays and array specific features (like SRDF). Physical storage arrays are highly optimized for their respectively targeted workloads and QoS goals.

It seems EMC is going to focus ViPR Data Services on capabilities that are value-add across array types, have low friction to inject, and present low risk to performance and availability. As a first shot, they are going to roll out HDFS/object storage as a service over file or block (or object), and are listing similar live data type "access/translations" possible in the future. The bigger vision is to enable new kinds of data constructs for new kinds of application workloads (and this ties into Pivotal's vision).

Overall, ViPR is a service provider "play", with multi-tenancy and global distribution, but the point is to enable enterprises and service providers to offer storage services to their clients without big adoption hurdles. ViPR in turn offers its own API for integration into newer and larger services that can benefit or leverage storage pooling, especially SDDC kinds of solutions. We will also see EMC reconstruct some high value storage features (like RecoverPoint) on top of ViPR.

ViPR looks to be the solution that marries the EMC portfolio into a holistic solution family - a neat evolution that shows that EMC continues to innovate hard.  We'll need a much more detailed look at to really appreciate how well ViPR delivers technically, but it now appears this time EMC's take on storage virtualization should be revenue-enhancing.

Because of the internal object nature, enterprise features like automatic snapshots on every write, inline deduplication, and scale-out clustering up to 6 nodes in a "ring" (with replication local across the "ring" and remote between "rings") are naturally included in the single low-cost solution. And those features operate automatically - no storage expert required.

There are a lot of very interesting features built into OneBlox - cloud-based management, full encryption, single global namespace, and what I would call BYODrive. You pop in your own drives that you source from wherever you want.  You can reuse, repurpose and recycle drives easily with OneBlox. To upgrade, just pull a drive out (data is replicated, not RAID, so no worries there), throw it away (its encrypted!), and push in a replacement (which will get data re-replicated back onto it).

Interested? Read more about Exablox, including two very interesting case studies (one with Museum of Science Boston and the other with Liquid Robotics) in the Taneja Group report, "High End Storage for Everyone". 

Since many if not most of today's "bigger data" problems are actually still in the TB sizes, this is really Big Data news.  We applaud the open source move, which we think will get Tokutek into many, many more conversation and projects than they could as a niche performance enhancer.

And for storage folks - did we mention that TokuDB has write-optimized compression?  That makes it a "flash" investment optimization too, for those accelerating database performance with flash hardware.

1. MMS combines previously separated processes: mobile device management (MDM) and mobile application management (MAM). This enables a central console, mobile utilities and IT policies for a vast swath of user devices.

1. The company has also added a highly secure Exchange email client for iOS and Android devices. The email client makes users happy with a full suite including email, calendar and contacts. It makes IT happy by granting centralized IT control for a major headache with BYOD: securing business email applications on user devices. ; T manages the suite using the MMS console.

1. MMS also enables secure single user sign-on for wrapped applications. Single sign-on is dearly loved by users but requires very strong access control lest multiple applications be compromised. Administrators can use policies and optional settings to control single sign-on rights and sessions. Authentication uses LDAP and SAML, and the process integrates with existing cloud security platforms from Symantec and CA.

Symantec is certainly not alone in the mobile management space: Cisco, IBM and others are also investing heavily. Symantec’s new announcement is interesting because they are combining MDM and MAM into a single console, which adds a welcome level of management and user efficiency, and broader security.

** We are finishing a landscape market report on enterprise file collaboration players. If you would like a copy please email christine.taylor@tanejagroup.com and we will send you one as soon as it’s complete.
 

While there are a quite a few speed optimization and streaming data packages being developed, hServer is built on proven production quality software. ScaleOut has a lot of field experience and hServer's grid is made to be not only elastically scalable, but also highly available.

As the name implies, this is a scale-out cluster "grid" solution. Unlike HDFS, when you add a node to hServer, it will automatically repartition and rebalance. Technically you can stream data into this grid in real-time and apply map reduce style computations in memory (actually something that their existing analytics solutions did before). In this way, total data motion is minimized. But since the world is Hadooping along, the new hServer brings some java interface libraries that enable Hadoop jobs (map reduce jobs on Hadoop) to acces the in memory grid while minimizing network overhead.  Since data is stored in memory in key-value pairs, I think of this as a type of converged noSQL/big data hardware solution.

In addition to enabling Hadoop to read/write from the in-memory grid, hServer comes with a library so you can quickly set it up as a read cache between Hadoop and HDFS. This won't require re-developing your Hadoop processing but could give you a big speedup. They report an 11x latency reduction over native HDFS reads.

To entice folks further, there is a free community edition for up to four nodes with no restrictions on using it in production, so now everyone can add an in-memory data grid to their big data platforms.  Definitely worth trying out if you want to develop a more operational edge to your computing.

• ECC is not knowledge sharing systems. These technology solutions are sometimes called “enterprise collaboration systems” and may include groupware, wikis, blogging platforms, distributed project mgmt, and web meeting software. They exist to improve human collaboration between remote individuals and teams. ECC might be included as a file sharing feature in a larger system but is not synonomous with group collaboration.

• ECC is not SaaS (Software as a Service) applications with collaborative features. There are hundreds of cloud-delivered business applications that enable data sharing such as Google Docs, salesforce.com or SAP’s online portfolio. For the most part their collaborative emphasis is extending its specific application data to mobile devices and remote offices. ECC evolves towards application file support but is not limited to a single SaaS application.

• ECC is not social media. Business social media links users and content but does not exist to synchronous shared files.

Now that we have discussed what enterprise cloud collaboration is not, let’s define what it is: Enterprise cloud collaboration (ECC) is a highly available and scalable cloud file sharing solution with centralized IT control.  

Note the four factors: 1) ECC must be highly available and scalable for enterprise-grade usage, 2) provides file sharing services from a public, private or hybrid cloud, 3) enables file sharing between multiple users and devices, and 4) offers a centralized console for IT security, user and data management.

But there are some benefits that come from market longevity, especially around SSD, and sTec has some differentiation, engineering, and bragging rights from their longevity.  There’s quite a bit of stuff that happens down at the device layer with solid state, and there can be a tremendous difference between vendors who have well engineered product and those who do not.  That’s why sTec initially had a stranglehold on the enterprise market - they could deliver sustainable, well-behaved, high IOPs performance in a resilient device.  With the increasing consumerization of SSD, we easily miss this difference, but there are plenty of war stories in the field from users who have been burned by poor devices delivered in arrays and/or servers.

So we think much of the market is looking for what adds up to real differences that they can stake their careers on, and may find sTec’s message pretty compelling.  The importance of engineering and innovation is part of the reason we’ve seen some interesting vendors acquired, think Anobit (Apple), Pliant (Seagate), SandForce (LSI), TMS (IBM), etc.  In fact, most of sTec’s competitors have been swallowed up, and they’ve lost quite a bit of their focus in that swallowing.  But with sTec, they’ve done this engineering themselves for a couple of decades now, and they can continue to deliver well engineered product without being subject to being swallowed up and redirected by a bigger company.  If you’re looking for a highly reputable device these days, it frankly might come down to sTec, Intel, or one of these other vendors who has been swallowed up by a bigger critter.  Which would you rather have?  For equivalent dollars, I’d put my money on the one with the most consistent focus and execution.

The challenges for sTec are three-fold.  One, there is a lot going on in the SSD market that can distract users from the fundamentals.  Two, there are a number of innovators who are crafting pretty slick software stacks that can compensate for the variability of underlying SSD.  NexGen, Kaminario, Whiptail, Nimble, Tegile, SimpliVity, Nutanix, etc.  All vary in maturity and capability, but “some” are doing a pretty sophisticated job of managing the underlying solid state to make it perform predictably, no matter the IO pattern, cycle time, etc..  Third, sTec’s diversification will require some great channel execution, and they need to empower that channel in some unique ways to make their approach competitive with some of the “already out there” vendors who are themselves giant marketing machines, as well as the tremendous competitive pressure from direct purchase.

Are there insurmountable hurdles here?  No.  I think the enterprise is listening for more details on the first two challenges, and sTec has a compelling and real story about the differences and where their engineering pays off.  That engineering can enable the enterprise to more flexibly deploy lower cost SSD storage anywhere they see fit - e.g. in the server.  In a currently running research project, we are seeing tremendous demand for just this flexibility - the enterprise customer clearly wants more and easier choice, and they want to be able to selectively deploy more accessible devices for specific demands.  More specifically, they have an eye on server deployments of SSD, and this is exactly where sTec technology could easily fit.  sTec’s timing may be great, if they can get and keep the attention of the enterprise.  These first two challenges are fundamentally marketing challenges.

The third challenge is all about channel program execution, and only time will tell.    But it is pretty interesting that of the challenges I see facing sTec, I don’t see technology challenges.  For sTec, it is strictly a matter of whether they can get their business machine tuned to their new marketing and channel program needs.  That is frankly, a pretty unusual spot for a technology vendor to be in, and it is quite possible this bodes well for sTec.

sTec may well grab your attention over the next year, and if you have a means to buy sTec storage for your performance needs, that will be all the recipe sTec needs here to be successful.  What do you think?  Do you see sTec technology in your future?

A couple of days ago, sTec finally formally announced their long rumored plans to change their go-to-market model and diversify from their traditional OEM focus.  In this shift, sTec has declared their intent to build a strong general enterprise and channel program that will carry sTec product directly into the hands of practitioners.

Squeezing value out of data is a competitive process. The more value you can extract faster, the better you can compete in your target market. This competition is what really drives recent trends to extract value out of the new "big data". Extracting value from big data was not cost-effective and often not even possible with common IT data processing architectures based on scale-up computing and relational databases.

Traditional BI solutions always required long lead times to develop ETL workflows and evolve data warehouses even over relational data. When looking at vast volumes and varieties of new big data, traditional BI approaches just can't keep up.   (As I was quoted in Justin Kern's Information Management article on Platfora)

Recent projects like Hadoop, fundamentally a commodity approach to scale-out computing and storage, have matured into commercially viable ways to unlock value from that "big data". Early adopters include web 2.0 companies and the largest of enterprises who have had both the budget and staffing depth to capitalize on the opportunity. Yet this leaves the vast majority of IT organizations to struggle with difficult functional programming (MapReduce) and large scale data management issues when they attempt to deliver big data based analytical services to their lines of business.
 
The potential value of Platfora is that it could instantly change the game of big data analysis for a wide swath of the IT enterprise market. Platfora leverages Hadoop's big data capabilities under the hood, layering on an intelligent and automatic self-service business intelligence analysis environment.

Platfora's secret weapon here is that they enable a business analyst to interactively explore, extend, and redefine their data interests (what Platfora calls a "lens") without requiring an explicit IT data model or data warehouse ETL workflow be developed or modified. Behind the scenes, Platfora works from the analysts interests to create and maintain an in-memory cache that is self-updating from the larger Hadoop data store. The result is a fast and agile solution that enables an incredibly interactive analysis capability. This would be impressive over a traditional analytical database, but over Hadoop it enables analyzing all the various volumes and varieties of big data.
 
Platfora promises to provide both analytical agility and effectiveness over big data. And help IT organizations of all sizes deliver high value data analysis services easily to their lines of business. We certainly hope that solutions like Platfora unlock incredible amounts of innovation and creativity that today is stifled by rigid and expensive BI approaches.

At the same time, IT and hosting providers must accomplish these availability feats cost-effectively for their business. The need becomes massively scaled storage with high throughput and capacity, with a high ROI on purchase and ongoing operations.

Required features for this type of long-term storage include the following:
•    Inexpensive power and cooling requirements.
•    Technology refresh capabilities that do not require data migration every few years.
•    Continuous data availability, as the point of long-term data retention on disk is immediate access.
•    Efficient provisioning management for large and growing data stores.
•    Massive scalability for both throughput and capacity.
•    Cost-effective capital and operational costs.

This is a tall order and NEC HYDRAstor 4000 is meeting it. The HYDRAstor’s architecture is grid-based for flexible scalability and massive growth. Its grid architecture enables dynamic auto-provisioning to create a single high-performance, high-capacity pool that is shared by file systems and applications. Its top model, the HS8-4000, exceeds 900 TB/hr throughput and grants up to 100PB of effective (not raw) capacity.

Its architecture is based on nodes that IT can add independently of each other depending on the need for throughput performance plus capacity, or capacity alone. The HYDRAstor can admit NEC Hybrid Nodes or Storage Nodes. The Hybrid Nodes increase both performance and capacity. They replace the older generation Accelerator Nodes that only increased performance, since NEC found that IT admins rarely scale throughput without also scaling capacity. The Storage Nodes efficiently increase capacity.

Another feature is that NEC officially supports up to three generations of nodes in the same system. This means that IT can upgrade a system node by node over time without doing a major technology refresh every few years. Customers can retire older nodes as needed and replace them or add to them with newer generation nodes. Before IT pulls the old node, the system will evacuate data and reassign the data.  IT then physically pulls out the old node and replaces it with new node, which the system immediately acknowledges and adds to the storage pool.

One of the real problems associated with long-term retention on disk is data protection. NEC provides a strongly resilient architecture with erasure-coding, dynamic resiliency levels for application priority, and no RAID overhead. The system reconstructs lost data as necessary without reconstructing healthy drives, and spreads the data across multiple spindles. Very high-performance global data dedupe operates inline.

NEC courts mid-sized companies and the enterprise with HYDRAstor. It is also active in selling to hosting providers. NEC partners with Iron Mountain to offer Cloud in a Vault, which provides cloud infrastructure and NEC products and services. The new release crams serious capacity into 4TB drives, which puts HYDRAstor 4000 into a more competitive price position with these customers.

NEC has good technology; that is not the question. The question is how they can compete effectively against deep-pocket competitors in the high end where HYDRAstor lives. We hope that NEC will hit and hit hard their competitive messages of speed, capacity, availability, efficiency, and cost-effectiveness. It’s a strong message that deserves to be heard.
 

 

Taneja Group’s just-completed survey of nearly 280 IT managers and executives – including a healthy mix of users and buyers – supports the market need for storage acceleration in general, and server-side storage acceleration solutions in particular. More than three-quarters of the respondents said that they are thinking about deploying a storage acceleration solution of some kind. More specifically, 56% of respondents indicated that server-based acceleration is one of the acceleration approaches they most value, with users overall valuing it more highly than array or network based approaches. But in contrast, only 26% said they had deployed a server-based acceleration solution to date, citing a number of obstacles that have prevented them from moving forward.

We believe that FabricCache adapters will help customers to overcome many of the limitations of today’s server-based caching solutions. Enterprise users have traditionally overprovisioned cache in single servers to ensure that critical applications can achieve high performance levels, even during periods of peak IO demand. This has been especially true in highly virtualized and consolidated infrastructures, in which multiple virtual servers or virtual desktops are contending for the same set of physical SAN resources. The QLE10000 eliminates the need for costly overprovisioning by allowing single-server flash cache to be transparently and coherently shared across multiple physical servers, irrespective of the operating systems, hypervisors and applications they’re running. In effect, this enables FabricCache to serve as a caching SAN adapter, the first of its kind in the industry.

FabricCache is also much simpler for IT administrators to deploy and manage than many of today’s server-side acceleration offerings. With FabricCache, customers do not need to install multiple drivers or a caching agent per OS; all that’s required is a single standard adapter driver, based on the QLogic driver stack that has been proven in use with millions of installed Fibre Channel ports. This is a particularly big deal in virtualized environments, in which virtual server acceleration solutions typically require a driver to be installed in each guest OS. Since the FabricCache adapter appears logically to the server as a single QLogic HBA, it can be integrated transparently and non-disruptively into existing enterprise environments.

One use case in which FabricCache stands out is in the live virtual machine (VM) migration between physical servers, which we saw QLogic recently demonstrate. In this demonstration, FabricCache adapters were used to cache SAN storage during a vMotion-enabled live migration (this migration took place in a VMware vSphere environment, but FabricCache supports other major hypervisors as well). Normally as the result of a vMotion operation, you would expect to see at least a temporary degradation in IO performance during and immediately after the migration, since the migrated VM needs time to re-warm its cache on the new server following the vMotion. But in this instance, FabricCache enabled the shared cache to stay warm, and performance remained high during the migration exercise.  And notably, advanced vSphere capabilities such as DRS workload balancing were not compromised. When you extrapolate the performance advantages from the two vMotions we saw in this demo to a large number of such migrations, you can see the potential productivity and quality of service benefits that FabricCache can deliver to virtual infrastructure users.

FabricCache also provides economic benefits: because it can be integrated non-disruptively into any storage and server infrastructure, running any combination of hypervisors, operating systems and applications, FabricCache eliminates the costs of additional drivers and caching software, and helps to extend the life of an existing SAN. We look forward to seeing how customers apply FabricCache in practice, as it begins shipping in the next few weeks.

At the same time, the results were surprising because every interviewee volunteered that Virtual Instruments is unequalled in its account relationships, customer support, and proactive engagement, especially compared to other systems and performance management vendors. This is a company that is doing something very right at multiple levels.

Digging a little deeper, we documented that VI has indeed helped these enterprises squeeze more out of their existing storage, enabled consolidation, and validated planned storage acquisitions. These companies offered multiple anecdotes about using VI to solve deep, highly visible, and potentially costly IO issues that they just couldn't address with any other solution, including with a storage vendor's own storage management. The recognized savings in both CAPEX and OPEX was real and easily justified an enterprise-wide deployment. In some cases VI's operational insight and optimization directly accelerates external competitive advantages in performance and service.

The report itself is a quick read about real people tackling real challenges. If you maintain storage in an enterprise data center or wonder what a great IT vendor does to deserve your patronage, you'll want to check it out - you can find it free here...

But it doesn't change some of the challenges for OpenStack, and may actually make some worse. OpenStack is no doubt flourishing, but as an open source initiative, a natural part of that process seems inevitably to be forks and the development of various different flavors. While OpenStack faces a number of challenges in developing beyond significant complexity, the biggest challenge for IBM may be OpenStack's current flavors and existing inertia in that direction. For a force as big as IBM, it will be challenging to get the troops to all march in line, and get the message across about a distinct OpenStack flavor and IBM differentiated capabilities. Among hundreds of thousands of developers who may already look to other sources for OpenStack, how is IBM supposed to say "this is the one to use". What will customers do if they engage in an IBM architecture and require services they can't get from other flavors of OpenStack, especially in large global organizations that might have not only the public cloud marketplace, but also several private clouds to choose from. And it can only be a short matter of time before IBM's flavor of OpenStack must look pretty unique - the current OpenStack contains considerable complexity that wouldn't make for a viable IBM private cloud architecture. As IBM takes up OpenStack, can it steer the platform well enough to suit its own purposes without breaking the openness of OpenStack? And can IBM tolerate the divergent nature (from where much innovation arises) of an open source platform when it becomes central to many products and their development ecosystem? I think only time will tell, as it isn't sufficient to argue that historic lessons from Linux provide an answer. Linux is a far different creature than a complex stack built to scale to data center sizes. Frankly, I was hoping that OpenStack would continue its rapid march forward through the efforts of innovators like PistonCloud, and at present, I think there is significant risk that this announcement draws away focus from the fundamentals, and suddenly transforms OpenStack into too big of an exercise for little voices to be heard. Can IBM propel maturity further forward without entirely consuming OpenStack into something proprietary?

Footnotes

As we look at the bigger market, IBM's OpenStack announcement has thrown down a strategy gauntlet with VMware - they are now both approaching the public cloud market with diametric strategies. One is trying to extend their hypervisor roots into a public cloud architecture, and foster an ecosystem of providers (VMware). The other is taking on cloud with an architecture that is first built for the public cloud, and turn it into a private cloud architecture, while likely standing up significant public services themselves (IBM).  Now you can use whatever terminology you want for two divergent sets of compute we see going on - systems of record versus systems of engagement for example - but the bottom line is both companies have a big stake in both types of compute, and neither is ready to give it up as an enterprise IT initiative they can enable.  VMware has a ton of resources focused on enabling new architectures for applications that are now moving to hyperscale environments in the public cloud.  IBM has built a tremendous portfolio that could be seriously hammered if IBM offerings were considered secondary to "where can I build this?" questions in new IT initiatives.

It is ironic to think that in deeper analysis, the histories of both companies might suggest that they would be naturally more inclined to the strategy of the other. These must be turbulent clouds to play this kind of havoc with strategies.

There will nonetheless be some interesting angles for a third major vendor to explore, especially if someone were to innovate in packaging and cross-cloud interoperability simultaneously.

When we talk to end users, and we hear frequently two drivers for virtualization. 1. Customers want to decrease their TCO, and virtualization can do it.  2. Customers want better IT capabilities, and many things can fall into this bucket.  When we survey customers, these two reasons are the top drivers for 90% or more of the respondents.

In terms of #2 above, the capability that customers most often identify as most important is business continuity.  But in this particular dimension, VMware doesn't always give customers what they expect out of the box.

Out of the box, VMware vSphere gives you great capabilities like vMotion, DRS, and ESXi clustering.  This can help pool resources, automate the placement of workloads, and make workload movement across a cluster pretty easy.  In turn, customers can move workloads around when they need to have planned outages to service or replace hardware.  With shared storage and in the event of unplanned outages, customers can also easily restart virtual servers on another system.  That's a pretty big improvement over the days of standalone servers, and can help make the business much more continuous.

Even better, for truly critical applications, VMware also offers an HA solution.  HA can monitor a virtual server, and when a heartbeat is no longer detected, can trigger the automatic restart of that virtual server.  If you're working in a cluster, that will happen on another hypervisor if there was a hardware failure.

But the reality is that VMware HA doesn't detect the majority of failures - application failures most often happen not because of a halted server, but because of application errors or server configuration errors.  Moreover, I've heard that VMware HA also restricts the use of some host settings (haven't actually investigated in the lab yet).  Most importantly though, VMware HA restarts can take a good 4 or 5 minutes.  These are drawbacks that may fail to meet your requirements for critical applications.

Over the past few weeks, I've been engaged with Symantec on a multi-fronted conversation around business continuity, and I've gone so far (along with Don Harrop) as to take a look one of their recent product offerings in the lab to validate their claims around how it delivers better availability.

So what did we find in our testing?

VCS speeds up failover events to the point that continuity is maintained. VCS patrols application integrity, and fully protects the viability of that application. Instead of running the risk of hung applications, and minimally tolerating 5 minute or more outages, a stateful app can resume with less than 45 seconds of outage, with very little overhead in exchange for the protection - and that's a key point - that very little overhead makes VCS efficient.

Now the final interesting angle I'll call out here is that Veritas Cluster Server actually speaks to the number one reason why customers pursue server virtualization too. If you remember what I said, I said this was TCO savings. The biggest factor in achieving TCO savings with virtualization is efficiency, achieved through the consolidation of workloads and higher VM density. Symantec has a message here too. The Symantec ApplicationHA and Veritas Cluster Server solutions deliver enhanced availability while sharing resources and enabling low utilization standby systems with less overhead than full clones running as standby servers, and other potential approaches.

VMware channel partners have all recently heard that virtualization comes with great demand for other complementary services, software, and hardware. Customers know this already. Well business continuity is one such area, and it is particularly important as virtualization begins to take on increasingly critical enterprise workloads.

So if you're looking for what other componentry you're going to need to bring along, you should turn some attention to DR, and as I've stated in the test report, take a look at Symantec's VCS as a clearly differentiated product. 

Full report is available here, or here.

Let me just break this down into a few key observations:

1. Convergence surfaced throughout the portfolio. For the first time in a long time I saw products throughout the portfolio - including networking - and they all consistently demonstrated convergence elements. Networking seemed to resurface with renewed vigor. This in itself is notable, as networking has been missing on my radar for a few years (other than integrated components like VirtualConnect and FlexFabric). Most interesting was the MSR30 and MSR50 systems that enable branch-in-a-box type deployments with blade embedded virtualization, even including a Riverbed Steelhead offering.

2. HP convergence is turning into more than just hardware - it is also innovative integration. HP has upped the ante in a far too under recognized way; perhaps under recognized because it is pretty subtle. HP is looking for integration that goes beyond hardware convergence. I already mentioned the virtualization enabled MSRs, but an absolutely compelling demonstration was the integration of VirtualConnect management into the vSphere administrative interface. VirtualConnect as many already know manages the dynamic provisioning of vNICs and vHBAs to blades in a C7000 BladeSystem. These ports are dynamically defined and connected to blades over the C7000's backplane and connectivity modules. With this coming integration, administrators will have some pretty mind blowing agility in dynamically defining their hardware connectivity in relation to what they're doing in the virtual infrastructure, effectively enabling a dynamically modifiable pseudo-backplane across hypervisors and between hypervisors and infrastructure systems like storage and the core network.

3. This combination is greater than the sum of its parts. As we start to look at what that last innovation does - VirtualConnect management within the virtualization layer - it starts to clearly define a new area of responsibility and capability within the data center. It enables server-based networking to an entirely new degree, yet simultaneously creates clear delineation between core networking and server-based networking for organizations with different functional teams in each area. Now combine this with a few other way under marketed capabilities, and some interesting thoughts ensue. Example one, HP Cloud Maps stand to automate the storage, network, and server configuration of entire sets of infrastructure. Example two, AutoFlex inside of CloudSystem is a demonstration of automated elasticity that can make physical infrastructure scale up and down as demands change. These types of integrations demonstrate an alignment between infrastructure systems and virtualization that can give even the emerging HyperConvergence players (Nutanix, SimpliVity, and Scale Computing) a run for their money. HP looks to be moving down a path that can weave multiple systems together (storage, networking, compute) into what for all practical purposes acts like a data center fabric and that serves up adjustable and adaptable amounts of these things as IT services. So why is that greater than the sum of the parts?

When you begin to think about a data center fabric, it is interesting to think about what services might be better off there. You could see some cases where it might be better to move specialized functionality out of the hypervisor for acceleration and performance, especially if those services can be equally agile in hardware. The HyperConvergence guys for example get this, and leverage scalable accelerated storage and data movement functions and high levels of automation. While VMware is pushing toward engulfing more functionality in the hypervisor for extreme agility via a software-defined data center (and including storage), it looks clear that HP hasn't abandoned the infrastructure to a future of commoditization.

There's an ideological battle brewing here, and HP is offering pretty compelling evidence that there's more to compute than software. The innovation is back, and for HP, it's clear they're going to enhance customer value through infrastructure. Since the 3PAR acquisition, I've never doubted HP storage had a bright future. But for the first time in a while, I see evidence across the entire systems portfolio that HP is going somewhere truly new, and it looks to have the potential of changing IT for the better. This may be dependent on sustainable execution on HP's part, and how well they can polish the resulting product that comes from these integrations, but for that, only time will tell. For now, I'm excited about their future.

I just tweeted a congrats to VMware for designing a Solution Provider to Service Provider incentivization program that in effect creates an App Store for buying public cloud infrastructure services. This program, called Cloud Credits, is a tremendous breakthrough for the industry, and it is subtle enough that many will miss the potential impact.  I have been pushing and waiting for just such a program for 2 years, and I have suspected that only HP was in a position to potentially do a program similar to this.  But that's another blog.

What is Cloud Credits?  Cloud Credits will be sold by the VMware channel, and incentivized like normal VMware software.  Using Cloud Credits, a customer will go to their My VMware portal and be able to select a certified vCloud service provider.  The service provider will then realize revenue from use of the Cloud Credits.  

The Cloud Credits are designed to keep the solution provider involved in the solution sale and implementation, and influential in the service provider relationship.  Service providers meanwhile can compete in this marketplace portal via pricing and their differentiation.

In my understanding, there looks to be future potential for solution providers to bring their own pass-through rebranded relationships into the portal, and the diversity of services may expand to also include things built on CloudFoundry - e.g., maybe it isn't an infrastructure service, but a DropBox like service built on CloudFoundry.

There's liable to be some growing pains as VMware figures out this mediator/brokerage role, but the incentivization model sounds well baked and compelling. The question is how fast can VMware chase a bigger market of cloud services that is much more diverse than Infrastructure as a Service (IAAS).

What's the big deal from all of this?

In a cloudifying world, how do vendors maintain customer relationships when they start looking toward the cloud for services?  There's a huge competitive threat to the traditional stickiness of a vendor, and the cloud in many ways stands to make everyone more disposable.  Meanwhile, successful cloud vendors have created on-going symbiotic relationships that are even stickier - whether you're considering Amazon or the Apple App Store.  Moreover, this stickiness often revolves around a digital community that is hard for a traditional vendor to match.  Key vendors in the infrastructure have a window of time to innovate here, while the cloud is still young, and innovation can fundamentally change the role they play.  There are profits to be had from the traditional solution provider, the service provider, and the act of creating a solution between them - if your product is central to all of those dimensions.  That is the nature of the App Store.

VMware has created something here that will be fundamental to their long term revenues and customer retention in a more competitive and diversified world of virtualization and cloud.  I believe they are also defining a model that many other vendors will follow, but many will be hard pressed to execute it with as much impact. 

The past couple of years have brought some interesting diversification to the Riverbed portfolio, and we continue to follow Riverbed closely.  Usually, I assume there is little from Riverbed that I'm not tracking and it is particularly notable when they surprise me.  Yesterday surprised me.

Riverbed over the past few years has moved from a highly specialized flavor of WAN Optimization, what we originally called WAFS, into an even comprehensive location abstraction of IT services.  They can effectively virtualize the location of an array of branch services today, allowing the business to keep consolidated and simplified services in their data center.  Such services today range from file to backup to VDI through products like Steelhead, Granite, and Whitewater, and these products can generally be physical or virtual.  

While Riverbed through acquisitions continues to move toward more comprehensive network services management that could someday turn into a comprehensive foundation beneath a wall-less software defined data center, a more tactical product caught my eye yesterday.

That product is a virtualized Steelhead installed on a blade in an HP MSR branch office box.  These MSR30 and MSR50 boxes from HP are converged multi-function devices that can be extended with a variety of modules that include compute blades.  

HP has the ability to move a lot of boxes.  If the HP field can understand this MSR box, and sell superior business value and the tremendous TCO advantages over traditional remote and branch office server infrastructure, then this could be something big.  The key will be helping the HP field understand the traditional Riverbed value and differentiation.  Could integration in HP channel be as big for RVBD in the mid-range as SRDF was in the enterprise?  The work is just beginning, and only time will tell.

..

jeff@tanejagroup.com

@JeffBoles

This is fully SQL compliant analytical database stuff, integrated with and powered by a tailored Hadoop implementation. Not only does this enable end business users to blast away with thier favorite existing tools and expertise, but the performance they get from this distro, as demonstrated today, is blisteringly fast over billions of records. Part of the sauce here is something they call Dynamic Pipelining (tm), which we will no doubt dive into as more details become available, but appears to be a data pipelining way to support parallel SQL queries with a more fabric-like intercommunication layer.  And when I say parallel queries, I should clarify that these are not limited to embarassingly parallel (i.e. simple) applications or queries.  HAWQ was shown to handle large multi-way joins, windowing functions, and other SQL analytical type queries with stellar performance.

This does bring up some philosophical questions about structured data v.s unstructured or semi-structured data, and the applications and infrastructures one can leverage to extract value from it all. This distro can handle both, and we expect to see some new application frontiers opened up. This kind of solution can be a game-changing technology as the normal IT enterprise comes to see Hadoop and Hadoop like solutions as the way all data could be processed in tomorrow's data center. Adoption of that vision may take awhile, but we think this HAWQ will take off immediately.

This claim is based on DDN delivering and supporting a fully integrated Hadoop appliance - complete integrated distro, full management, compute servers, networking, and their Storage Fusion Architecture storage.  By leveraging SFA (and a little RDMA magic) they can deliver massive IOPs faster than DAS meaning they can process huge volumes of data with a smaller, and importantly an overall cheaper, cluster. And all that data is in enterprise class storage - ready for any kind of big data workflows. This is one optimized platform, and worth a much deeper look, especially if you are in the market for a enterprise class Hadoop solution of maybe more than 100 nodes and don't have or want a large staff of open source wiz kids to implement and maintain it.  SI's with vertical markets will definitely want to pick up on this as a platform and layer in a targeted solution as a complete package.

This morning at VMware PEX, my presentation focused on not just what VM density is, but why it matters to IT customers.  The simple fact is that VM density makes a tremendous difference in the cost of ownership for any given set of virtualized infrastructure.  In the virtualized data center, a big portion of customer costs revolve around the physical server and hypervisor.  The more applications or workloads you have per physical server, the lower your costs will be.  There will be fewer physical servers, and therefore less hardware, licensing, power and cooling, and on-going management.  If you think about your costs distributed across all of those virtualized applications, higher VM density will lead you to lower cost per application, and optimizing that cost per application should be the bottom line focus of the IT manager.

Then we spun through a hands-on test exercise from Taneja Group Labs, and looked at how 3PAR stood out when compared to traditional but pretty big mid-range system.

Here's the presentation.

..

jeff@tanejagroup.com

@JeffBoles

As I'm sitting here at VMware's Partner Exchange, and in HP Boot Camp sessions, I just digested a refreshing perspective on VDI initiatives (Client Virtualization in HP parlance).  HP, unsurprisingly, messages VDI as a services and consulting undertaking, with a specific methodology and specific HP engagement tools like CVAM - HP's Client Virtualization Analysis and Modeling.  This is reality - VDI is an architecture and design effort.  But marketeers might have you believe otherwise.

How often do you see a marketing spiel that talks about a single dimension product, and how it delivers VDI at a break through cost per desktop.  It isn't that simple, and there's a lot more architecting to do in the background.  Sure, storage is a big challenge.  Sure, fast storage makes a big difference and can make the other parts of a VDI project easier, but there are actually few companies out there that have a product that can really solve the master set of VDI architecture complexity.

Now the few that can offer more than a single dimension fix have some real value that should stand out.  HyperConverged companies like Nutanix and SimpliVity, and even V3 (who we haven't recognized as formally HyperConverged) can arguably go farther in simplifying the activity of design and architecture for VDI environments.

For most, it is an architecture exercise, and it should be missed that the fundamentals are in understanding the requirements and getting the design right.

As full disclosure, HP covered my travel costs in exchange for participating in their client virtualization track.

..

jeff@tanejagroup.com

@JeffBoles

I just arrived at PEX last night, and I'm looking forward to what might unfold this year.  I've spent the year waiting to see where VMware Partner programs are headed in a cloudy, on-prem/off-prem world, and whether they'll take a leadership role in defining what collaborative partnerships look like in the VMware ecosystem.  Meanwhile, I'm poised on the edge of my seat to see what steps VMware takes to further define their "vision" of Software Defined in terms of real product functionality and advanced orchestration.

Off of this main topic, I'll be hunting service provider challenges as they build ever bigger infrastructures.  Who's the leader at VMware PEX in terms of service provider storage infrastructure?  My bet is on HP 3PAR storage going in - after all, they defined the utility storage concept that is now core to the service provider pursuing XaaS.  We'll see what informal interactions with attendees turn up.  Are you a service provider?  Maybe we'll run into each other.

..

jeff.boles@tanejagroup.com

@JeffBoles

Up until now, mid-sized VMware customers looking for a bundled operations management solution have had the choice of vCenter Operations Management Foundation (vC Ops Foundation), a free-add on for customers running vSphere standalone licenses; or the vCloud Suite Advanced, which bundles vC Ops functionality. This portfolio left a significant gap for mid-sized organizations (typically in the range of 500-1000 employees).

vSOM – which bundles vC Ops Standard edition – fills that gap, building on vC Ops Foundation to deliver more advanced performance analysis capabilities, including Dynamic Thresholds and Root Cause Analysis. Another example is vSphere Change Detection and Performance Correlation, which will automatically alert an administrator if the behavior of an object changes. vSOM also provides advanced capacity optimization functionality, including capacity metering, modeling and right-sizing features. vSOM is priced per processor without limits on core, vRAM, or number of VMs.

By making vC Ops Standard capabilities more accessible and affordable, vSOM will enable mid-sized customers to more productively manage a greater number of VMs while maintaining quality of service levels, and thereby get more value out of their vSphere installations. Kit Colbert’s recent post provides more details and puts all of this into context: http://cto.vmware.com/bringing-together-vsphere-and-operations-management/.