Join Newsletter
Forgot
password?
Register
Trusted Business Advisors, Expert Technology Analysts

Taneja Blog

Taneja Blog / Data Center Systems

EMC announces PCIe Flash Cache—Fusion IO gets its first major competitor

In another demonstration of genuine leadership and street fighting spirit EMC announced today the availability of VFCache, otherwise known as Project Lightning. Simply put, VFCache is a 300GB SLC-flash on a PCIe card, designed to go into a server which is running an IO starved application, perhaps even after using the fastest HDDs and even array-based SSDs. Such applications are more prevalent than one might think. In fact, more and more applications have become IO-starved over the past decade as CPUs and networks have become faster and storage has improved marginally, if at all. The product is also GA at the same time.

EMC has recognized for years that to compete effectively against companies like HP, IBM and Dell, which sell complete solutions, it must deliver server-based technologies that impact storage. PowerPath is a great example of that, as is VFCache. EMC cannot let its system supplier competitors gain an edge just because they sell servers and it doesn't. Buying VMware was a fantastic example of how EMC stole the show against these systems suppliers. Delivering industry's second PCIe flash cache is yet another example of such bold ferocity. Fusion IO has been enjoying uncontested growth in this space for almost two years. Until now. EMC just one-upped them by a large margin. To be sure, there are other PCIe flash-cache cards in the market. For instance, Marvell has its DragonFly HBA that allows scalability by using up to 8 PCIe cards in a cluster for some very impressive performance numbers. We believe, however, at least for now, the major competition for EMC will come from Fusion IO.

The key difference between Fusion IO and EMC's VFCache is that VFCache offloads flash management tasks from the CPU - things like wear leveling, garbage collection, and such. These are heavy users of the CPU that by all rights should be dedicated to serving the application. According to EMC, for a 100% read applications (worst case scenario), Fusion IO solution could use up 20% of the CPU vs. 5% for EMC. When a customer had no alternative in the market any product that boosted application performance was a welcome relief. But giving up even 20% of the CPU is troublesome for most applications. If this difference between the two products pans out, it alone will make a huge dent in the sales of Fusion IO cards. Granted there is a built-in bias on the part of HP, IBM, Dell and other system providers to use a non-EMC product. So I am sure they will continue to suggest Fusion IO to their customers (HP has a deal to resell Fusion IO cards) but let's not forget that EMC sold storage into those same accounts to become the largest purveyor of storage. So let's not underestimate its ability to sell large quantities of VFCache into these servers, even if the storage is non-EMC. And for EMC's storage-only competitors (NetApp, HDS and others) the gap further widened.

But for now EMC will probably focus on its existing accounts and the lowest hanging fruit. We can safely say that we are not aware of a single enterprise account where there aren’t 10s, if not 100s of applications that are starved for IO. The opportunity is immense, especially given that the solution is transparent to an application and the customer doesn't have to do anything except decide which LUNs (and therefore, which applications) should be under the control of VFCache. All software interactions are hands-free and require no operator intervention. Note that VFCache only deals with reads and uses the write-through method for writes. Applications that are predominately write-oriented (backup, for instance) are not candidates for VFCache.

The caching game just changed big time. EMC continues to amaze us in terms of identifying new market opportunities and taking the lead by getting there first, relative to its main competitors. Either by developing a product like VFCache itself or buying the company with an innovative product before others do. In this case, they saved a few billion dollars (for not having to buy Fusion IO) and yet gained a serious advantage over their main competitors. Game on!

  • Premiered: 02/06/12
  • Author: Arun Taneja
Topic(s): VFCache EMC Fusion IO SSD Solid State Devices Application Acceleration cache

Comments

There are no comments to display. Scroll down to leave your own!

 

Leave a Comment

You must be logged in to comment. Click here to log in or register if you don't have an account.