EMC NetWorker 8 launched in June 2012. I’d just spent 5 hours recording a video for the launch. Customers were excited about the new architecture and the tighter integration with Data Domain. Sales were already hyped by the revenue growth started with NetWorker 7.6.2. The NetWorker 8 launch was going to be an “I’m Back” moment. Between the adrenaline and the caffeine, I was vibrating as I walked through the building. Then a NetWorker engineer sidled up to me and asked, “So, does this launch mean we’re killing NetWorker?”
This week’s question – “Are we killing XtremIO?”
Portfolio Companies – Nothing Ever Dies
Companies decide who they are going to be: consumer vs. business, product vs. services, profitable vs. unicorn. One of the biggest choices is whether to be single product vs. portfolio. At a single product company, there is little confusion about what product matters, but you can get constrained by the limits of that product. At a portfolio company, you can build/buy whatever you need, but there is complexity in having multiple products. And, of course, at a portfolio company, you have to answer the “are you killing Product X” question. As an old CEO said, “That’s the life we’ve chosen”.
Over the past few years, each product has had at least one customer question whether we’re killing it. My favorite was the customer who asked if Centera would kill Data Domain. (Yes, I did write that sentence in the proper order.) The truth is, it’s almost impossible to kill any product at EMC. Each product has at least one massive customer who has built a business-critical process around it; that customer, of course, will stop working with EMC if we don’t support that product. At a portfolio company, there are no “independent” product sales. That’s why we’re so careful about adding a new product – once you’ve put it in, it’s there forever.
The result – it’s almost unheard of to kill a product, much less a high-growth, high-revenue product.
All-Flash Storage is not One-Size-Fits-All
If you think of “All-Flash Storage” as a market, like “Purpose Built Backup Appliance” (think Data Domain), then it makes sense that you think you only need one product. If all that matters is the storage media, then the functionality, cost/performance, reliability, availability, and protocols are irrelevant. The applications will be so thrilled to have “All-Flash Storage” that they’ll re-architect to fit whatever limitations the storage system has. Not only does the universe revolve around storage, but it revolves around storage media. (Not surprisingly , narcissism is often an attribute of a single-product company). That’s the logical conclusion of the arguments I hear.
All-Flash Storage is not a new market. Flash storage is disrupting the storage media market, but not the storage market. All-Flash has become ubiquitous across both traditional and new arrays and vendors, so it’s important, but it’s not a separate market. All-Flash does drive architecture and product evolution, but we’re all still building storage arrays. If storage media changes really did create new markets, then I want more hype around Barium-Ferrite tape and shingled magnetic recording disk (SMR). (Note: I really don’t. On the other hand, the DNA storage is cool.)
What Makes Each All-Flash Product Special
Each of the all-flash primary storage products in the Core Technologies portfolio (VMAX, XtremIO, Unity) has features and architectures that make them unique. (Disclaimer: With the amount of engineering invested in each product, there are thousands of “favorite architectural choices”. These are mine. If you want to tout others, feel free to use the comments section or spray paint them on the side of my house; my HOA already hates me, anyway.)
VMAX – Caching/Tiering. Everybody knows about the performance, reliability, availability, data protection (SRDF, SnapVX, ProtectPoint), protocols, etc. At the core, however, the FAST (Fully Automated Storage Tiering) algorithms fire me up. When everything is “All-Flash”, though, who needs caching/tiering? In the next couple of years, however, we’ll see persistent memory, flash, shingled disk, and cloud storage – understanding how to best store and serve data will matter more than ever. VMAX has the metadata analytics to be the storage media broker for core and critical applications.
XtremIO – Versions (Copy Data Management). When people think of XtremIO, they think of speed and deduplication. Storage speed and space efficiency isn’t enough to survive in the modern metadata center, though. XtremIO’s dedupe is just the first way to expose the value of the block sharing architecture. The sustained value comes from creating and distributing lightweight copies for test & development, data protection, and analytics. Unlike many systems, I don’t need to worry about complex links to other copies (block sharing creates independence), the performance hit on the production copy (scale-out), or crushing the network (dedupe-aware data movement). XtremIO has the metadata management to be the system of choice for the DevOps and analytics world…
Unity – Simplicity. Most storage systems are skewed toward simplifying enterprise data center challenges. That means scale-out, heterogeneous storage managed by dedicated IT teams. I love scale-out (see VMAX and XtremIO), but it complicates install and management of well-known consolidated workloads. Similarly, I love heterogeneous data protection, but sometimes replicating between two similar systems is better (especially if one is All-Flash and the other is Hybrid to reduce costs). I see the value in feature-rich dedicated storage management, but sometimes I want to just get a basic environment up and running in 15 minutes or less. As the metadata center shifts toward wanting simple, agile, application/VM-driven storage, Unity has the simplicity to deliver.
Each system has a design center that any “one-size-fits-all” product can’t deliver. Of course, there are many workloads that all of these systems (and many competitor’s products) could handle without a problem. There will always be a baseline of functionality (e.g. in CDM, caching/tiering, and simplicity) that we’ll deliver across the platforms, and even strive to drive the baseline higher. In virtually every environment, however, key applications will drive requirements to optimize in some direction. I believe that these 3 design centers will be the most critical in the storage space.
EMC is a portfolio company. The storage products in our portfolio will continue to evolve with the media transitions. Remember, All-Flash Storage isn’t a market – it’s a point in time for the type of media we’ll use in our systems. Soon enough, you’ll hear about All-Persistent-Memory Storage. Regardless of the media, our storage systems are differentiated by their software design center, and you’ll see us continue to extend our functionality in those areas. Each system will be part of the Modern Metadata Center.
In summary, when we talk all-flash: XtremIO is not dead. VMAX is not dead. Unity is not dead.
Oh, and in case anybody is wondering, NetWorker is alive and well, too.
Stephen Manley @makitadremel