For weeks the brain trust at Wikibon and I have been dissecting EMC’s V-Max announcement and we still have a ways to go. Our technical guys want to go even deeper and we need to put together a good roadmap to help clients go from where they are today to this new vision of the virtual data center– if that’s where they really want to go.
Why Wouldn’t Customers Want to Migrate?
Well for one thing, it will require much planning. And between VMware, Cisco and now EMC everyone needs to figure out how much of this stuff is actionable versus white board fodder. It will also be disruptive. Big architectural changes like this that require forklift upgrades are like the scene in the Godfather when Michael asks Clemenza “How bad it’s going to be?” and Clemenza says it will be pretty bleepin’ bad. “Probably all the other Families will line up against us. That’s all right. These things gotta happen every five years or so…Helps to get rid of the bad blood.”
It was just around five years ago that EMC shot the original bus-based Symmetrix and moved customers to the current DMX architecture. EMC had let down its guard back then and scrambled to catch up to the technology curve. This time feels different– EMC understands it’s more profitable to keep customers than to lose them and try to win them back. The company is actually getting pretty good at this game and from what I can tell is working hard to preserve existing processes and scripts built up around Symmetrix. The better job EMC does here the more loyalty it builds and customer loyalty is pretty high in EMC’s Symmetrix base. Just spend a couple of days at EMC World talking to Symmetrix customers and you’ll confirm this fact.
EMC’s competitors and the anti-EMC blogosphere are invoking the typical bromide around how this product announcement is a yawner and a copycat of other architectures. And how performance is unclear because EMC won’t publish benchmarks, etc. But for one thing I haven’t seen a storage announcement get this much attention in a while, maybe ever; and secondly, has EMC ever failed the performance test when it comes to Symmetrix? I’ve very rarely heard a customer say EMC’s products don’t perform in real world applications. Hiccups, yes. Broad-based failures to deliver high service levels? Not a widespread problem.
The fact is the main criticisms of Symmetrix have been that its hardware, software and maintenance are incredibly expensive, moving stuff around is cumbersome and requires lots of rocket scientists, migrations are complex and need to be planned carefully and there’s no integration with other EMC platforms (e.g. CLARiiON). Yet customers keep buying the stuff because there’s value there.
Does V-Max Address these Shortcomings?
Yes and no. If you want to play devil’s advocate about this announcement, let’s first focus on the economics of tiered storage. Can a flash/FC/SATA V-Max architecture or even a flash/SATA combo compete with Tier 2 modular storage from the likes of Compellent, 3PAR, HP/Lefthand, Dell/Equal and ‘cheap ‘n deep’ storage architectures like Nexsan? And can a virtualized EMC V-Max infrastructure match the simplicity of say a Compellent system? The answer is it doesn’t have to head-to-head, rather EMC must get close enough and demonstrate that from an OPEX standpoint tiering in a V-Max infrastructure will be less expensive and/or provide greater benefits than stove-piping across disparate architectures.
EMC’s pre-announcement of Fully Automated Storage Tiering (FAST) is an admission that without FAST the answer is essentially no, EMC can’t compete on cost. With FAST, assuming it’s all EMC promises, this will be very interesting and I expect many customers to re-think their current ‘Tier 1 Avoidance’ strategies. If for example EMC can deliver Compellent-like automated tiering at V-Max scale, look out. If EMC could deliver that today this announcement would be ridiculously unfair.
The other big gap is CLARiiON. How does that play in this architecture? Does it integrate? Does it remain a silo? So far it looks like it doesn’t play and while EMC’s silence on this question is kind of a bummer, its unclear that integrating CLARiiON is a reasonable expectation in the near term.
Finally, can EMC really make an N-way clustered storage infrastructure look like a single system that shares memory, ports, interconnects and other resources? I honestly don’t know at this point and that’s an area the Wikibon technical folks will be dissecting over the coming months. I expect a few customers will be too.
How Should Customers Approach V-Max Negotiations?
Interestingly, some of the commentary I’ve read suggests that given the economic downturn, this is a terrible time for EMC to be making such an announcement because they’re asking customers to absorb incremental infrastructure costs to exploit the V-Max architecture. I couldn’t disagree more. It’s early in the year, demand is horrible (Seagate just announced demand for enterprise drives is off 33% year on year) and this transition will take time. EMC can hold customers’ hands through the downturn, offering sweet deals on DMX-4′s to clear out an inventory that is already being carefully managed. EMC services can sell point projects to plan for the transition and get some proof of concept activity going– it’s perfect timing leading into the slow summer season.
Now is the time for EMC customers to enter the planning phase and utilize internal or EMC services to plan for the future. Part of that planning should include establishing a baseline of efficiency measuring factors such as utilization and migration timeframes. Customers should work with EMC to set targets for how V-Max will improve these metrics and clients should hold EMC’s feet to the fire applying assumptions around the impacts of Virtual Provisioning, flash, FAST, tiering, energy efficiency and the Virtual Matrix architecture in general. The advantage of at least partially using EMC services to perform this work is to gain agreement and secure efficiency guarantees which can be plugged directly into ROI calculations and utilized for future negotiations.
To optimize these strategies, users will have to aggressively execute multiple levels of engagement over the next 3-6 months, at least. This is particularly critical as EMC has pre-announced FAST which can be used as leverage to negotiate forward efficiency gains and pricing plans in near-to-mid-term contracts.
Organizationally, storage, application and network operations, finance, legal, and procurement will play important roles in these negotiations. To achieve the best results, these groups will have to be as coordinated as the EMC sales and service organizations. Treating negotiations as a project can force establishment of clear objectives, roles, negotiation guidelines and a range of desirable outcomes.
Given the ongoing growth rates of storage , the continuous introduction of new function, and the high switching costs between platform products, ensuring attractive forward-looking maintenance contracts is at least as important as agreements that focus solely on acquisition costs. EMC may likely be in a mood to negotiate more aggressively in this regard and customers should explore EMC’s willingness to re-assess existing maintenance contracts as part of new V-Max negotiations.
While the V-Max announcement doesn’t create a procurement imperative (i.e. buy now or miss out), EMC contracts set up over the next several quarters will shape the scope of storage negotiations for some time.
Just make sure you steer clear of offers you can’t refuse.