Mike Workman, CEO, Pillar Data Systems

Jeff HawkinsThe idea that a storage hardware start-up could succeed
against entrenched players like IBM, EMC and Hitachi Data Systems seemed
almost laughable a few years ago.


The hardware sector slogged to a crawl after corporations were chastened by
the reality of the ever-shrinking budgets.


But Pillar Data Systems, a company created by Oracle CEO Larry Ellison in
2001, kept plugging away. Last summer, the startup launched
a system that offers block-based, storage area network (SAN)
and file-oriented, network-attached storage (NAS) in one
machine.

The idea of offering a device with the ability to handle SAN and NAS
protocols is nothing new. Storage vendors, such as EMC and Network
Appliance, have been putting together machines that don’t discriminate
against SAN or NAS transmissions for a few years.


But Pillar’s Axiom 500 server uses a “peered NAS and SAN architecture,”
pooling both NAS and SAN in one repository. The approach seems to be
working.


The San Jose, Calif., company is close to securing its 100th customer and
will launch a smaller version of the Axiom 500 this summer, CEO Mike Workman
told internetnews.com recently.


Q: Did you find it hard to crack into a market where there are so many major
hardware vendors, such as IBM, EMC and Hitachi Data Systems, already
entrenched?


It’s not a cakewalk. People don’t just have you come in and eat their lunch
and stay for dinner. They’re definitely fighting us. Axiom 500 covers a
broad spectrum. It goes from tier 3 to tier 1 applications and customers’
offices. It does block storage and file storage. The price range is $75,000
to over $1 million. We have a whole table showing where our one platform
displaces our competitors’ products today.


We displace the [EMC] Clariion CX500, the CX700, the [NetApp] NS500 and the
NS700. With a single platform, you can cover such a broad spectrum of
storage problems or opportunities in a customer’s data center. They only
have to know one thing. They don’t have to work with two vendors. They don’t
have to have three different management platforms, etc.


Q: Talk about the tiering you do in the box to give data different
priorities.


There has to be different quality of service in storage.
Today, if you request storage bytes or blocks from a SAN LUN, it doesn’t
know the difference between you and somebody else.


So you would want the storage to treat requests based on the performance
requirements, bandwidth requirements, and your ability to contend for those
resources. You’d like it to be a little more intelligent than first come,
first serve, which is what most storage is today.

What Pillar did was say:
“We’re going to define classes of service, just like in a plane. They get
first-class seats. They’re bigger, they have less people per cab. They have
more flight attendants, better food, and entertainment.”

What are the
analogies to storage? It’s just not where the data sits on the disk, which,
in the case of first class, data sits at the outside of the disk. We always
put the first-class data, the first-class seats, near the window.


What Pillar does is give you a utility player. It’s something that, by
software configuration, solves just a panoramic scope of problems. We’re not
everything to everybody. There are certain things you’re better off buying a
point solution for. Instead of it being a silo, we solve 85 percent of the
problems instead of 15 percent of the problems.


Q: And one of the value propositions is that the storage admins you sell to
don’t have to have both NAS and SAN training to use the Axiom 500.


If you want to look at one of the things that’s really disruptive about the
box, it’s that NAS is easy. Network Appliance used to say it’s like a
toaster. What we did for SAN is make it the same. It looks that easy and it
is that easy, which was hard because SAN’s a little more complex. It’s
always hardest when you make things look easier than they are.

Q: With the Axiom 500 doing so well, how do you follow that act?


We have a new product, the Axiom 300, coming out this summer, which will cut
our entry price in half. It’s the same platform as the Axiom 500, uses the
same disk storage as the Axiom 500.

In other words you can take it off one
and put it on another. It has the same management interfaces, so if you know
how to run the Axiom 500, you know how to run the Axiom 300.

The difference
is that we carved out about half of the infrastructure of the Axiom 500 to
offer a better entry product for the one to 40 terabyte range. What’s great
about it is you can take that system and upgrade it to the Axiom 500.


Q: So, customers said: “Wow this Axiom 500 is great, could you do something
smaller?”


Yes. It was: “We liked this. But we don’t want to pay that much and we don’t
need that big of a system. Will you sell us one for $45,000?”

And the answer
is, that’s a hard price range to get with our system, so we built one for
them that puts that in the cards. The thing is built so that in its most
modest configuration, you have a data controller, two storage shelves and
pilot with 20 spindles on it.

What we decided to do was build a system that
starts a little bit smaller, say with 10 spindles, and that allows you only
to scale to 100 spindles. The Axiom 500 goes from 20 to 864.


Q: Were there challenges for the engineers to make a smaller model?


Yes. Basically, what we did is said: “Okay, we have twice as much processing
power as you need with that many spindles.” And so we took out some of
those, and we took out some of the fiber infrastructure and put iSCSI
on because iSCSI isn’t exactly making its way into the big-time enterprise where the Axiom 500 is. We’ll start with iSCSI in the Axiom
300 and move up.


Q: How do customers settle on Pillar when it comes to the competition?


The big customers in many cases have kept their shop simple with a single
vendor. In some cases, vendors don’t offer all tiers of storage, or if they
do it’s multiple platforms and it gets costly just to manage them.

When we
go in, we offer the ability to tier with one product. We also offer the
cost-effectiveness of tier 3 that spreads into tier 2 and tier 1. People
aren’t used to that. Small customers can’t afford tier 1, a DMX or a
TagmaStore, but they can get the kind of performance and availability and
flexibility out of a Pillar System that the big customers can offer with
more expensive platinum iron.

News Around the Web