DATAllegro heads for the high end
DATAllegro Stuart Frost called in for a prebriefing/feedback/consulting session. (I love advising my DBMS vendor clients on how to beat each other’s brains in. This was even more fun in the 1990s, when combat was generally more aggressive. Those were also the days when somebody would change jobs to an arch-rival and immediately explain how everything they’d told me before was utterly false …)
While I had Stuart on the phone, I did manage to extract some stuff I’m at liberty to use immediately. Here are the highlights:
- DATAllegro’s market sweet spot is 50-100+ terabyte databases. They have very few sales under 50 terabytes. (At various points in the conversation he said the sweet spot was 50+ terabytes, said the sweet sport was 100+ terabytes, and talked about 15-20 terabyte marts. This is very typical vendor behavior — their bigger, flashier customers are naturally foremost in their minds – but it does make things challenging for us analysts. I think the way I phrased it to begin this bullet point is probably the most accurate.)
- One disclosed example is 30 + 120 terabytes, with 200 terabytes more being installed over the next few weeks. That’s a telecom company, segregating billing records by age (and hence frequency of query).
- DATAllegro has “tens” of customers, with average deal size of $3 ½ – 4 million.
- DATAllegro’s quarterly revenue is “a little less” than Netezza’s was 12-15 months ago. I’m not sure exactly what that means, since Netezza happened to have a dip in the April 30, 2006 quarter to $12 million, vs. $17.7 million or so in both the quarter right before and the quarter right after that, as per the Netezza S-1.
- DATAllegro’s growth rate is a “lot” more than Netezza’s traditional 50%ish range.
- Most of DATAllegro’s sales are to Teradata shops.
- Stuart agrees – as do I – with the general consensus that data warehouse appliances have been used mainly for departmental/business unit data marts … but stay tuned.
- Slightly contrary to what I previously thought, DATAllegro is putting encryption cards into the Dell servers they now ship, and this is actually an important feature. Stuart notes that their compression of 3-3.5X (new in the recent release) helps with encryption performance. Gross I/O rates are actually down from the last release – which is what matters for encryption throughput — even as net I/O rates are up (both by roughly a factor of 2).
- Stuart was on his hobbyhorse about Netezza reliability, asserting that’s a reason most Netezza sites are well under 50 terabytes. (The idea is that if there’s a high failure rate of disks, then a big system with lots of disks will have so many failures as to drive you batty.) Frankly, this was one of the least convincing parts of the call, although it does seem likely that the EMC boxes DATAllegro now uses are more reliable, disk-for-disk, than Netezza ones.
- DATAllegro claims not to have lost any deals to Greenplum. A big reason – supposedly — is that since Greenplum has a more conventional indexing approach than other appliance vendors, it has trouble with near-real-time data loading, even though otherwise it scales well.
Keep getting great research about data management and related technologies. Get a FREE subscription by RSS/Atom or e-mail!
Technorati Tags: DATAllegro, Netezza, Teradata, Greenplum, data warehouse appliance
Comments
4 Responses to “DATAllegro heads for the high end”
Leave a Reply
I’d suggest that:
There’s a big reality gap between the (growing but reasonably limited number of) 100+ TB installations the appliance vendors claim to be focusing on and the real market opportunity which is something like 1.5 TB to 20 TBs which is where the mixed workload RDBMS vendors run out of steam (becoming very expensive in terms of specialized hardware and and support staff) and the point that the tried and tested Teradata platform becomes a compelling option. I’m sure everyone likes to talk about giant Wal-Mart sized installations but that doesn’t mean that everyone is Wal-Mart.
Sun/Greenplum is currently a compelling online data storage platform. They may or may not be be a good analytical platform but their crucial issue is a lack of public DWH/BI ecosystem support which is required to get them over the line in tenders and similar competitive situations- they need Informatica, BO, Cognos, MicroStrategy and the like to provide platform support to be a serious competitor.
Disk failures? Is there something special about DATallegro’s drives, RAID levels, engineering or the way that the platform stresses the disk in comparison to anyone else?
Alex,
If you’re suggesting that the Oracle-substitution market where Netezza mainly plays is much bigger than the Teradata-substitution market where DATAllegro mainly plays, you’re surely right. Especially in terms of units, of course — but notwithstanding Teradata’s hefty revenue, you’d even seem to be right in terms of dollars.
And your comment about third-party software support is spot-on.
As for disk reliability — surely there’s a reason why EMC is in business despite not being the price leader, and specifically a reason other than buyer stupidity.
Best,
CAM
Curt,
Thanks for posting the update.
In an attempt to clarify our ‘sweet spot’ here’s what I thought I said:
– Our sweet spot is 50TB and up.
– Beyond 100TB, we feel we are the only viable vendor other than Teradata (which is far more expensive). Note that Netezza’s range maxes out at 100TB.
– However, we do still compete for smaller deals – especially when it’s for an initial project within a large account.
Since overall MTBF is a factor of the number of components involved in the system, reliability is greatly impacted by the number of disks used for a given TB capacity. In a Netezza 10800 with 100TB capacity, there are 896 disks. In a 100TB DATAllegro appliance, there are no more than 192 disks (even using our lowest capacity racks), so the MTBF of the disk arrays is automatically better. In addition, the disks are higher quality (EMC certified) and hot-swappable. The EMC units we use were recently independently assesed as having five-nines reliability.
I won’t comment on your revenue estimates – got to leave something for you analysts to analyze!
Stuart
Alex,
You’re obviously correct in saying there are more installations in the 1.5TB to 20TB range than in the 100TB+ category. However, we’ve been surprised at how many opportunities we’ve been able to find. It’s certainly keeping us very busy!
Also, disruptive vendors have to start in a niche where they have compelling differentiation. Boiling the ocean is simply not going to work. So, while our product can be used for smaller DWs, we’re happy to focus on the high-end for now.
As for disk reliability, see my comments above. Also, we can categorically state that the EMC-certified drives we are using now are FAR more reliable than the WD ‘enterprise class’ drives we used in previous versions. I believe the same WD drives are still used by some of our competitors.
Aside from the obvious advantages of having EMC’s brand and salesforce behind us, improving reliability was the key reason for switching to EMC.
Stuart