iSCSI SAN Comparisons - Part 1 of n

10/29/2009

People across the industry are weighing on what the ‘best’ SAN or iSCSI SAN on the market is. It’s a perpetual question in any competitive market, especially a growing one like iSCSI storage. And iSCSI is growing, make no mistake, especially in comparison to other storage types. An Enterprise Strategy Group (ESG) survey of 500 enterprise IT managers from about a year ago said, “Forty percent report that they have already deployed iSCSI storage systems, and 27% plan on doing so in the next 12 months.” (Excerpted from an ESG TCO study on EqualLogic that cites the "ESG 2008 Enterprise Storage Systems Survey, November 2008”). That was a year ago and if you add all the new deployments Dell, HP, EMC and NetApp have done with iSCSI since then, I’d bet we’re not far off of a combined 67% of ESG’s representative sample.

Point being, iSCSI is here, real and deployed. Giving credit where credit is due, iSCSI’s growth isn’t just due to EqualLogic and Dell, but also the success of other vendors, HP/Lefthand included. iSCSI is in the boring part of the ‘hype cycle’ (my words, not Gartner’s), where you don’t hear about it in the press as much, but it’s being adopted in gobs, by 10s of thousands of different organizations, ranging from a couple terabytes at a 12-person hedge fund in New York to petabytes by government and intelligence agencies I very sadly can’t name.

To be clear, I’m with Dell’s storage product group in product marketing, and spend most of my time working with the EqualLogic line. I see some customers evaluating LeftHand as another iSCSI SAN option, but I understand that most (or the market plurality anyway) tend to vote with their wallet for Dell EqualLogic, which market share statistics from two major analysts firms point out. Beyond what the market indicates, IMHO the Dell EqualLogic solution is technically superior to LeftHand, but then again I’m a little biased, as is any other commentator whose bread is buttered by a vendor partner or employer.

A couple recent posts (here and here for example) in the blogosphere have started comparisons with LeftHand, with some incorrect information, and I’d like to clarify a couple things. This isn’t meant to be an exhaustive or fully scientific study on comparisons, but rather clearing some brush from road signs that competitors may have obscured.

-Clustering –Thinking EqualLogic and LeftHand have ‘clustered’ SANs. It’s a convenient somewhat lazy, but not accurate basis for comparison. EqualLogic doesn’t actually ‘cluster’ as LeftHand does, using multiple non-redundant servers/nodes to provide redundancy. EqualLogic scales out performance and capacity with multiple arrays, but each array is fundamentally redundant and designed for 5-9s, much like HP’s recommended enterprise arrays the EVA and XP. While some may thinking SMBs don’t need or want storage that is considered ‘enterprise’, I think SMBs would prefer to have arrays that have enterprise reliability and features if they can afford it.

Also, I think it’s fair to say that SMBs consider their data critical and want enterprise-class protection when consolidating onto an advanced SAN. Part of enterprise-class protection is being able to keep safely handling updates to critical data even if a controller is offline. Purpose-built storage from Dell, or offerings higher in HP’s portfolio than LeftHand, can meet that goal without needing to make a second copy. Some people think that with LeftHand they’re meeting that goal with a second copy, but they’d really need three copies according to research of Lefthand’s user manual; so there’s a chance many aren’t following the HP LeftHand user-manual best practices for protecting critical data but don’t even know it.

-Groups – Some people misunderstand the EqualLogic peer storage architecture, but I’d invite you and them to learn more. First, data doesn’t span PS Series SAN groups, it can be replicated between them, and it can be replicated between a lot more than 2 of them. Again to provide system redundancy, EqualLogic doesn’t require multi-group clustering as LeftHand does. You can have many more than 2 arrays within a group and data volumes span and are load balanced across multiple arrays within a group. SMB customers for instance may choose to start with a SAN comprised of 2 PS4000 entry/midrange arrays, but they can always expand that group with additional EqualLogic PS6000 arrays.

-Cost – some people make the assumption that because LeftHand products use server hardware that EqualLogic is more expensive than LeftHand. It might sound reasonable as a high level statement, but it’s another lazy, subjective statement at best. If a customer factors in true TCO, the total power and cooling need for the n nodes need for LeftHand redundancy and the amount of disk needed to reach a given performance target for a workload, I suspect the true cost picture is much different and favors EqualLogic.

There are many more reasons to consider Dell EqualLogic over HP LeftHand, but I’ll stop for today and summarize to say that if you follow best practices and consider business-relevant risks, you’ll see that EqualLogic has advantage for SMBs and enterprises, as well as being a simpler, cleaner and denser solution. Based on some other FUD that’s floating out there in the market, I suspect we’ll have to add another chapter to this post, so stay tuned.

-- Dylan Locsin, with Rob Young at Dell --

Comments are welcome and can be posted on the thread below.