A StorageMojo reader has a problem. Can you help?

Our mail hub (80,000+ mailboxes) is virtualized with vSphere 4.1 with Red Hat Enterprise Linux 5 x64 and Dovecot 2.0 [an open source IMAP/POP3 email server for Linux/UNIX-like systems]. We are using HP LeftHand Networks P4300 iSCSI storage in a “network RAID10 setup of RAID10 storage” for Dovecot indexes and multiple “networks RAID1 of RAID5 storage” for actual mailboxes.

This is my take: our Dovecot indexes are getting hammered with lots of small I/O requests, about 8,000 IOPS continuous during 8-working-hour days, 75% write. Indexes are fairly small (50 GB) and expected to grow to 100-150 GB, but need a lot of random I/O. We need real-time replication in storage (LeftHand is ok for us) and we think that SSD should shine in this situation. Bandwidth is not a problem (200-300 megabits of indexes traffic, but we need more IOPs).

The problem is the indexes, but our total mailbox capacity is expected to grow to 6 TB compressed using zlib compression in Dovecot.

We want to buy a storage appliance with the following requirements:

• Vsphere 4.1 & 5 certified storage, VAAI enabled (if possible)
• iSCSI (1 gbps)
• High number of IOPS (at least 12,000+, most of them writes)
• Small size (200 GB)
• Fault tolerant (RAID, battery-backed write cache, power supply, fans, multiple gigabit uplinks, synchronous replication)
• Cheap (less than $30k the full setup)

We want to buy at the beginning of 2012. Any product that fits?

The StorageMojo take
Suspect price will be the most significant limiter. But the respondent only needs index storage not the whole shooting match. He’s pretty happy with LeftHand for mailbox storage.

But if we can solve both problems for him, why not? If he should relax some constraint, feel free to suggest it.

He’ll be watching the comments, so if you have questions please ask them. I’ll be following the comments as well.

Courteous comments welcome, of course. His email was edited for clarity.

A shout out to all ‘worlders: the StorageMojo analyst team will be descending en masse on the lush Nevada desert (yes, in Phoenix real estate people often tout the “lush desert” – don’t they know what desert means?) to take in our 1st VMworld.

We’re arriving the 29th & departing the 31st. Hope to learn a lot more about storage in the virtual world.

The StorageMojo take
Data storage is hard enough when your CPUs aren’t VMotioning across the globe. Dealing with the related issues of concurrency, consistency and limited network bandwidth makes it even more challenging.

I’m interested in seeing what Virsto is announcing as well as what is coming from companies I haven’t yet heard of. Do readers have any suggestions for vendors that shouldn’t be missed?

Courteous comments welcome, of course. Even though it is a hike to the Venetian and the Wynn, I’m going to stay in one of the new City Center hotels.

Robert Gray, a former analyst at IDC and long time friend, asked me to respond to a comment from NetApp president Tom Georgens.

Tom told Information Week that:

. . . virtualization is fundamentally changing the way data centers are designed: instead of individual infrastructure around individual apps, they’re building one single infrastructure that’s independent of the apps. And now they want the same with storage because customers are seeing that virtualization is the way to dramatically improve the flexibility of the business as well as IT costs.

(bolding added)

NetApp’s on a tear
NetApp is growing, approaching a $5 billion annual run rate. Mr. Georgens attributes part of that growth to the 50% capacity savings guarantee that NetApp announced two years ago.

That guarantee, which originally applied only to the storage of VMs, appears to have morphed into a more generic talking point: buy NetApp and we’ll give you the same capacity with half the boxes.

Gee, a marketing initiative had an impact on the bottom line? Don’t tell HDS!

The StorageMojo take
It is natural to conflate the impact of virtualization and the growth of single infrastructure thinking. But that is an accident of timing, not cause-and-effect.

The single infrastructure meme grows out of the obvious success of scale-out processing and storage architectures, as exemplified by Amazon and Google. Even without VMware the economic advantages of Internet scale cloud computing and storage would drive the industry.

What virtualization is doing is driving IT architects and CIOs to rethink long-settled – fossilized – infrastructure practices. As they do so it becomes clear that the standard stove-piped enterprise IT model no longer makes sense either economically or operationally.

True, virtual machines are making the desirability of a common storage infrastructure obvious to even the most hidebound IT shop: migrating storage with a VM is tough enough – why have a different infrastructure at the other end? But virtualization or no, the old model of bespoke infrastructure for each application was not economic. Today’s glass houses are factories, not artisanal workshops, and they need industrial scale infrastructure.

This builds on the long-term secular trend of the commoditization of IT. This is why all the array vendors have been busy migrating their custom hardware to software running on x86 commodity boards.

Ironically, NetApp doesn’t actually offer a single scale-out infrastructure, with the exception of recently acquired Bycast. The Spinnaker acquisition – lo, these many years past – was supposed to solve the problem but the integration issues have proven more difficult than NetApp ever imagined.

Enterprises will never reach the scale of a Google or Amazon. However, they will find that scale, even at the enterprise level, has its own dynamics. One implication is a widespread move to object storage.

More on that later.

Courteous comments welcome, of course. Robert, I hadn’t seen Tom’s words. Thanks for bring them to my attention.

Got a presentation from a company I hadn’t heard of before: VKernel. Their specialty is virtual machine capacity planning and management.

They demo’d their software, which ships as a VM. They offer free versions that are good enough to tell you if their paid tools will be able to help you.

VKernel has a customer sysadmin in the room, talking about how his company – which remains nameless, but has over 1,000 VMs on 40 C-class blades – uses the product. The user says that some of the big wins have been in reducing over-provisioning and quickly finding performance problems.

The user says that it is very lightweight and agentless. It installs in 15 minutes or less and quickly provides actionable data, even to junior admins.

Today the product supports VWware and their Capacity Optimizer also supports Hyper-V.

The StorageMojo take
Storage planning is a lost art: many admins grew up in a 1-app/1-server world where capacity management meant adding another drive to a server. Virtual machines are much more complicated.

There’s the I/O blender effect, which renders 40 years of I/O empiricism useless. The interactions between VMs on a single physical server. Contention for memory, bandwidth and IOPS.

It’s a new world. VKernel looks promising.

Courteous comments welcome, of course. This presentation is part of the Gestalt IT blogger tour that VKernal, among others, helped pay for. If you’ve tried VKernel please comment on them below. Update: And I think I’ve finally learned how to spell kernel. Thanks Joe, Stephen. End update.