Atrato: High-performance, high-density storage

by Robin Harris on Tuesday, 12 February, 2008

Got an interesting press release this morning about a Denver-area company, Atrato, announcing its existence and $18 million in funding. Their mission:

Based in Westminster, Colorado, Atrato Inc.’s (www.atrato-Inc.com) mission is to help companies in entertainment, the Web, IPTV, HPC and VOD open up infinite new worlds of content for customers by offering them high-speed, high volume data access. Atrato’s high-density storage system with integrated data acceleration does nothing less than change the economics of high-speed/high-volume I/O processing.

So what do they have?
They say very little about their technology in the release:

. . . breakthrough technology, a high-performance storage platform that is designed to eliminate the barriers to high-speed / high-volume data access, unlocking revenue and opportunities for a range of applications and industries.

The web site makes some more specific claims which are excerpted below:

Speeds to support any load level. Easily handles traffic spikes with the power of hundreds of servers energizing your site.

Content is protected at both the stream and hardware levels (in flight and at rest) to ensure the security and integrity of your content while the sealed array eliminates most physical security vulnerabilities.

The industry’s only three year maintenance-free, fail-in-place operation available today and has been granted hundreds of patent claims with numerous others applied for.

. . . up to 10,000 I/Os per second or 3000 streams in 5RU . . . .

[bolding added]

How do they do it?
They say little about the secret sauce, but after looking their web site and some patent applications I’ll venture this much:

  • The core team is heavy on hardware guys. With the numbers they’re quoting this is an ASIC-enabled box – think BlueArc for I/O. Lots of internal parallelism, wide stripes and mirroring.
  • They’ve developed some innovative packaging technology for high-density disk – I’m guessing up to 400 2.5″ drives per enclosure. You can’t easily replace the drives, so they’ve made a virtue of necessity and “sealed” the enclosure.
  • Beyond the high-density packaging they’ve thought long and hard about how to ensure a 3 year operational life. Offset counter-rotating drive pairs to damp rotational and actuator vibration, high-flow cooling and ample hot-spare provisioning are key.

The StorageMojo take
Way cool! Hardware is cheap, labor and downtime expensive so their architecture works from a TCO perspective. Sticking these boxes in cable system head-ends will simplify content distribution and support at the same time.

The prices are likely to look high, but when you factor in the 3 year maintenance contract it should be persuasive. With 80,000 IOPS from a single 42U rack it may even find favor in more I/O intensive environments.

This is the kind of innovative packaging I would have expected from Xyratex. Congrats to the Atrato team for a thorough re-thinking of storage infrastructure.

Comments welcome, as always. Atrato team members?

{ 2 trackbacks }

Startups Try to Speed Up Storage - GigaOM
Thursday, 14 February, 2008 at 1:39 pm
Atrato: a new kind of high performance storage « Storage Effect
Wednesday, 20 February, 2008 at 7:32 am

{ 7 comments… read them below or add one }

Gary Watson Thursday, 14 February, 2008 at 10:49 am

The herringbone stuff in the patent applications is interesting, but as far as mounting pairs of drives in counter-rotating back-to-back couplets for RV cancellation — this is something we started shipping in our ATABeast back in 2002. It continues to this day in our SATABeast.
Gary

xfer_rdy Saturday, 16 February, 2008 at 10:57 pm

They look like they are competing with 5 year old technology. Their specs maybe ok for teleco systems, like voice mail…

Until recently, I believe this company was called Sherwood Information Systems or something like that. They’ve had a face lift.

The world must have really gone crazy at some point…. 10,000 transactions per second… Uhhh IOPS ? On their web site they call they it “tough stuff”. This is 2008 !!! if you can’t do 120,000 IOPs in a 5RU give it up.

Simultaneous streams…. some minor bandwidth and caching 3000 HD streams 2 mbs (bits) —- 750 MB/s (bytes). Its a real screamer !!! You can get that performance with $400 raid controller, an 2gb/s fibre channel card and a crummy PC.

Counter rotation ?? There is some debate on whether its snake oil, that is unless the disk resonances are tightly coupled (that’s usually called a poor enclosure design). Others swear by it..

Ok, since we can’t see it, and they claim up to 80% saving on rack density and power … In comparison to what ?? 9gb drives ?

90% saving on connectivity ? Yes, if you are talking about T1s in comparison to fibre channel.

Reliability and down time is critical for this market. The carriers and operators are concerned with annual cost per stream. Today, cable operator headends are crammed with equipment with limited power, so they may have good chance in that market. As soon as operators can dump that analog equipment (Feb 17, 2009), racks space and power will be available and the door will close on that niche market.

SUN has an enterprise product they sell for $50/stream retail. They can be competitive.

Hopefully the Atrato management team will have the relationships in place to close some deals.

With the emerging business of on-site leased storage into carriers and operators, for pennies/gb (

val Wednesday, 20 February, 2008 at 12:58 pm

Simultaneous streams…. some minor bandwidth and caching 3000 HD streams 2 mbs (bits) —- 750 MB/s (bytes). Its a real screamer !!! You can get that performance with $400 raid controller, an 2gb/s fibre channel card and a crummy PC.

How do you propose to get 750 MB/s with a 2 Gb/s FC card ? 2 Gb/s = 200 Mb/s, at best !

As to 120K IOPs, the number is realistic only with RAID cache hits, i.e. reading from memory. When you go to the actual disks (as soon as there is no cache left), the number is closer to 15-20K IOPs. Write IOPs are naturally even slower, about 40K when hitting the cahe, about 10K when hitting the disk. Also, the numbers of 20K read IOPs/10K write IOPs were achieved with about 120 disks and four 2Gb FC ports. (http://bbs.doit.com.cn/attachment.php?aid=5055)

So, I don’t quite understand your scorn…

Mad Scientist! Sunday, 24 February, 2008 at 10:24 pm

Val is quite right on both points.
If the streaming could really be done with a $400 RAID card and 2GbFC on a crummy PC then there would not be a market for this. The aforementioned configuration would certainly work for 1 or two streams but not 3,000 streams – that is an entirely different problem that researchers have been struggling with for at least 15 years – I know this because I am one of those researchers.

The 10K IOPS is probably a worst-case scenario for purely random small read operations over 100% of the capacity of the Atrato array where cache is useless. Given any sort of locality of reference, the IOPS would go up significantly. I think they could easily meet or exceed the numbers that Val mentions.

Pretty amazing stuff if you ask me.

CDB Sunday, 24 February, 2008 at 10:44 pm

3000 SD MPEG-2 streams @ 3.75Mb/s is well within the bandwidth available to the Atrato system and, at 5RU, is very competitive with VOD offerings currently on the market (whether flash or disk based). The interesting thing that remains to be seen is which VOD vendor will be the first to announce pairing with a system which not only offers a solid content vault but the ability to increase overall streaming capacity?

As far as HD @ 2Mb/s, well, it’s not likely you’ll see that in a broadcast plant any time soon. I know there are some online streaming services that claim they can deliver streaming HD in that bandwidth, but I doubt any of us will be linking our PC to the HDTV to watch it. Realistic HD levels range from extremely compressed (MPEG-4) 5Mbps to standard rates such as MPEG-2 20Mbps.

Steve Monday, 17 March, 2008 at 2:47 pm

Hmm wonder if in 2 years IBM will buy them??

Steve Kowal Friday, 16 March, 2012 at 8:42 am

While it may be true that comparisons of INDIVIDUAL performance characteristics between this 2006 technology (the year I worked on the “SAID” design at Sherwood Information Partners) and today’s offerings may seem unimposing, when observed as an overall solution to the multifaceted problems faced by large “data storage farms” it becomes quite impressive.

The combination of Eric Wendel’s insight (primary founder of Sherwood and creator of the “Sealed Array of Individual Drives” concept) into the big picture of problems faced by mass data storage providers, and the pure depth of practical controller design innovation provided by Dr. Sam Siewert, resulted in a solution that excelled on all fronts simultaneously. Eric’s study into the low cost commercial 2.5″ spindle (when architected in a unique pseudo-parallel array configuration) as a means of providing a new and powerful solution in terms of power consumption / cost / data “accessage” (a term coined by Eric) coupled with Sam’s ability to transition concept into reality provided an overall solution to the entire IOPs/Bandwidth/Power/Cost/Size/Enterprise Reliability equation that many – if not most – still don’t grasp in terms of completeness and elegance today.

Leave a Comment

Previous post:

Next post: