There sure seem to be a lot of inaccuracies on ParAccel’s website
In what is actually an interesting post on database compression, ParAccel CTO Barry Zane threw in
Anyone who has met with us knows ParAccel shies away from hype.
But like many things ParAccel says, that is not true.
Edit (October, 2010): Like other posts I’ve linked to from Barry Zane’s blog, that one seems to be gone, with the URL redirecting elsewhere on ParAccel’s website.
The latest whoppers came in the form of several customers ParAccel listed on its website who hadn’t actually bought ParAccel’s DBMS, nor even decided to do so. It is fairly common to to claim a customer win, then retract the claim due to lack of permission to disclose. But that’s not what happened in these cases. Based on emails helpfully shared by a ParAccel competitor competing in some of those accounts, it seems clear that ParAccel actually posted fabricated claims of customer wins.
Another thing that was both technically and substantively false was ParAccel’s claim to be CERTIFIED price-performance leader. Obviously, this was meant to give the impression that ParAccel had been “certified” as the leader in price/performance, when the closest thing to that that was remotely true was that ParAccel had a leading position in the category of “price/performance measurements that happen to have a certification process.” At least, that was true for a short time; then ParAccel’s certification was found to have been erroneous, and got revoked, which did not however inspire ParAccel to immediately take the claim off the front page of its website.
ParAccel’s website also reflects a lot of praise from flagship customer LatiNode. What it perhaps understandably neglects to mention is that LatiNode is in a dormant state, placed there by acquirer Elandia due to LatiNode’s criminally corrupt customer acquisition practices.
I also don’t believe ParAccel’s endlessly-repeated claim that is has never lost a benchmark on performance. However, I must in fairness note that while I’ve been given names of customers who are supposed counterexamples to this claim by somebody I trust, I’ve never been able to actually verify those supposed ParAccel losses.
Comments
24 Responses to “There sure seem to be a lot of inaccuracies on ParAccel’s website”
Leave a Reply
Mr. Monash, I am an employee of ParAccel, and while we normally refrain from posting to your site as it would only bring further attention to your inaccuracies about us, this time you have gone too far. My company removed 3 logos recently. I can tell you for a fact that all 3 are ParAccel customers as of Q4’09…. I have seen the actual contracts and I think you might even have to agree with me that a contract from a public company trumps an email from a whiny competitor. I even worked on one of these deals. As you suggest sometime happens, I believe these three customers contacted us and asked that we remove their logos as it was not their policy to allow their vendors to post them. We complied
I suspect that your information comes from our closest competitor who wants us to look bad because we beat them. Actually, it makes you look bad when you do not verify facts before writing about them. You have our contact informatoin. I would encourage you to assume more responsibility for validating facts before you write about them. Oh, and by the way, I can also tell you that we worked hard on the LatiNode deal and delivered substantial improvements to their operations before their demise…, which had nothing to do with our DW project, as you can probably imagine.
-Lokesh
I know of one case where a mutual NDA was signed for a POC, and someone at ParAccel confused it with an actual sale and listed the potential customer as an actual customer on their website. ParAccel removed it as soon as the mistake was pointed out to them. I don’t know what happened in the other cases, but in this instance it was NOT a deliberately fabricated claim. Keep in mind that this “ParAccel competitor” who “shared” these emails with you has their own agenda. Did you question this person as to whether the sharing of these emails might violate any NDA’s his/her company might have with their potential customers?
What did ParAccel did so bad that now they are on your watch list? 🙂
Curt,
Thanks for the kind words on the compression posting. It is meant to be educational and informative in a straightforward manner.
However, and you knew this was coming, your additional points are incorrect:
We do not post fabricated claims of customer wins. Please forward to us the customer names our competitor claims are false. We will request that each customer send you confirmation. In exchange, you must then post the name of the competitor that is making such outrageous claims. Fair?
ParAccel’s TPC-H 30TB benchmark was published and certified to be the price-performance leader by approximately 16X. The benchmark was challenged and withdrawn about three months later. The TPC proceedings are confidential to TPC members, so I cannot provide details, but suffice it to say that ParAccel would have re-published, in a TPC-conformant manner, with the same performance if the equipment was available. However, as your September post assumes, Sun’s imminent acquisition by Oracle had made obtaining the equipment impossible.
As for Latinode’s bribes in Honduras, this is news to us. The people we dealt with were honest and honorable and they purchased a ParAccel license in 2007.
We have never lost a customer benchmark or bake-off on performance. We support your right to disbelieve, it makes the conversation more lively!
Lokesh,
Logo errors happen. Contracts get signed on evaluation deals that are not recognizable deals, and logos get put up anyway. It happens.
But when 20%+ of your logos get taken down the same week — well, that’s unusual.
My best take is that one of them is a deal you may still actually win, one is a deal you’ve been bragging about for years without any compelling evidence of deployment (certainly your competitors also in that account can’t find evidence that you are there), and I don’t know about the third.
Barry,
Thanks for posting!
As previously established, we have sharp disagreeements the significance of TPC results. Meanwhile, I note with amusement that your promised three-part series of blog posts about them never got beyond Part 1. 😉
But even if you’d been able to refile, or never had your results withdrawn, I would have found that home page phrasing to be deeply misleading, to the point of dishonesty.
Also — if you’re so proud about the lack of tuning needed to get great performance, why is it necessary for you to refile with the same (no longer cooperative) vendor? Why can’t you quickly throw something together on somebody else’s equipment?
Finally, please tell me, if you can, that the story I’m hearing about you doing a customer benchmark by putting different parts of the database into memory before each query and only counting processing time after it got there are just incorrect/lies/a big misunderstanding …
Thanks!
CAM
Not just Honduras. Yemen too.
And while it doesn’t invalidate your claim to have helped them achieve whatever business success they did, it does cast doubt on the magnitude of that business success. The company was sold for around $25 million, and $20 million of that is estimated to have been fraudulent, because of the bribes. Helping build a business worth $5 million is not nothing, but it’s also not as impressive as it could be.
Lokesh,
As for your comment about contact information — you have some nerve. Kim Stanick told me she’s under orders not to talk to me. She even BLOCKED me on Twitter, which given how the system works is an extreme move. Before doing so, she made an inappropriate comment about NDAs.
ParAccel has quite a ways to go before it deserves any benefit of the doubt.
Maybe that will change at some point, or has already changed as signified by ParAccel people posting in this thread. But that was the situation as known to me at the time of my posting.
Apologies, gentle reader, in advance for the length of this response, there are several points. Please refer to Curt’s and my previous text for the context of these phrases:
1. Customer verification
2. TPCH references
3. In-memory operation
4. Latinode
On customer verification, I repeat my challenge, which you haven’t responded to. Please forward to us the customer names our competitor claims are false. We will request that each customer send you confirmation. In exchange, you must then post the name of the competitor that is making such outrageous claims. Fair?
On TPCH, when we published our 30TB results, it included a full disclosure report of everything done. It was audited by a TPC-certified auditor, of which I believe only four such companies in the world are certified. The schema and all loading and query times were included, as well as the scripts that produced them. This was done on a publicly available forum, accessible to friends and foes alike at tpc.org.
Publishing ties up a cluster for at least a week. The 30TB benchmark was submitted to the auditor less than three weeks after the hardware was powered up. We ran the same database software available to customers, and ran without indexes or materialized views. We have chosen to publish with the sponsorship of a hardware vendor. There is a simple political expediency that hardware vendors need to decide whether they wish to risk the wrath of legacy database companies, that goes with publishing with us. With the imminent consolidations in the industry, I think more hardware vendors will be so inclined.
On in-memory operation, I think you must be confusing us with another vendor. In terms of customer benchmarks, we have meticulously run on-disk unless the customer asked for in-memory operation. We include a utility command to clear the memory cache between queries to guarantee this. I don’t know whether any other vendor goes to this extreme. Partly to remove any chance of confusion, we now have the disk cache disabled in our default configuration file. This way, if a customer does their own POC and is unaware of the utility command, they’ll see the same performance across runs. The disk cache is very beneficial, and the customer should enable it, only when the bulk of their queries use the same tables/columns.
To be clear, ParAccel supports both in-memory and on-disk operation. Fully in-memory operation is simply a matter of specifying a large “disk cache” and configuring for memory preload during database initialization. As was Oracle’s 2009 TPCH-1TB, our 2007 TPCH runs were done in-memory. That is no secret. The ParAccel 2009 TPCH 30TB was not done in-memory, all scanning was against the disks. To preempt any confusion, please note that this is a “scan cache” not a “results cache”, two altogether different animals.
Per Latinode, I can’t speculate on their actual value, I’ll only observe that some companies like Bear Stearns and Lehman Brothers were considered worthless and liquidated, while others that appear equally fragile are worth tens of billions of dollars in the public stock markets. Plus, in 2007, everybody thought the telcos were cowboys and the banks were stable. Good thing I invested in the automobile industry. Go figure. And to be exact, it was Honduran and Yemeni bribes… at least!
Curt,
Your understanding does not match the fact pattern. 3 logos were taken down, they are all Q4 customers with established bookings,as far as I know we had never messaged that they were a “customer” until now.
-Lokesh
Lokesh,
Your alleged customer relationship with Sears — that other incumbents can’t find evidence of — had been mentioned privately for a loooong time.
Barry,
I don’t understand the logic of your “challenge”. If the companies are willing to say, even in private reference checks, that they’re really customers, you’re in great shape. If they aren’t, then you have a problem.
Maybe you’re doing a good job of selling to multiple departments of the same enterprise at once, where the left hand (denying a sale) doesn’t know what the right one (acknowledging a sale) is doing. If so, good for you.
If not — well, your marketing is too focused on claims of customer wins for it to be excusable if a significant fraction are bogus.
Anyhow, as per my post title, my main point is that ParAccel’s website (and its marketing claims in general) are remarkably unreliable, as per specific examples of unarguable inaccuracy or radical difference of opinion. An increasing number of industry observers seem to agree that ParAccel’s are basically not to be trusted.
That said, it’s certainly possible that some subset of the claimed customers are actual ongoing users who are actually happy with a ParAccel system that is behaving not too dissimilarly from how a POC/benchmark suggested it would. I can’t find any quotes from same on ParAccel’s website, nor have I ever heard a convincing story of one (LatiNode excepted, back when they were in business) — but maybe they exist anyhow.
Curt,
I am not at liberty to say whether we were ever engaged with the company you mention. Either way (booked customer or merely engaged prospect), a vendor’s activities with a prospect are customarily private.
If we posted the name in January and then took it down, they are a signed, contracted customers who asked us to do so. We did not post logos of any non-customers. If you want further validation, I am told we will provide it under NDA offline, but only if you agree to publish a recanting of your claim that we did anything misleading in this regard. I do not believe we have ever deliberately mislead anyone, it would be very much against our culture to do so. I know you enjoy saying otherwise, but your claims have always come without hard evidence and appear more personally vindictive than journalistic.
-Lokesh
Curt,
Regarding your recent complaint about ParAccel’s customer logo postings, I am to blame for putting our newest customer’s logos on the website before authorized (they are indeed newly signed customers). I took the logos down within minutes of being requested to do so. The premature posting of those customer logos was nothing more than normally-acceptable marketing aggression, but this time I moved too soon.
Lokesh, Barry,
My tolerance for ParAccel trying to dictate the content of my blog posts was used up years ago. Please just stop with that theme.
If you want people to speak and write kindly about you, then disclose what’s true, add no more than a standard amount of spin to it, and don’t make emphatic claims that aren’t true.
Speaking personally, I most particularly never want to hear again from ParAccel “What you learned previously is no longer true, so we want you to ‘confess’ that it never was true in the first place.” I’ve gotten way too much of that over the years from you guys; I found it deeply dishonest when you first did it to me years ago; and my opinion hasn’t changed.
Thank you,
CAM
Hi Kim,
So of the customers whose logos survive on your site, which actually represent current production deployments of ParAccel? (As opposed to, say, no longer active deployments, deployments that haven’t happened yet, or resellers who have hope of future deployments.)
Anyhow, Barry, let’s simplify the challenge — I challenge you to prove that the companies you claimed were customers really are. Fair enough?
And Kim — all other issues aside, I’m glad your boss is finally allowing you to post comments here. 🙂 I hope you continue doing so in the future.
Curt,
Regarding your earlier comment: We list customers on our web site. Companies listed there chose to buy PADB, either on its merits alone or after competitive evaluation. The status of their deployment isn’t indicated by the listing. Further information about a listed (or unlisted) customer is shared privately in references for new prospects. If we have something further to say publicly about a customer, you’ll see it in a case-study on our web site, or in the press.
Kim
Kim,
By “chose to buy”, do you always mean that they signed a contract that includes non-trivial revenue to you, recognizable (immediately or at least by the time you post the info) under Generally Accepted Accounting Principles? (GAAP)
I ask because an awful lot of apparent customer wins these days seem to really amount to final evaluations, in which one has to wait a quarter or two to see whether they turn into solid sales or whether the product gets returned w/o obligation.
E.g., I have a client who seems (based on another product they bought) to have a penchant for signing a contract first and doing a final evaluation second. Well, Vertica didn’t put up their logo until months after the initial deal got signed. That strikes me as a much more responsible way to proceed.
Ok Monash. I stumbled on your website and found this thread quite amusing. Very clever that you’ve now have drawn in employees of Paraccel to discuss your lunatic blogging. But I’d like to submit that we get all of these DB companies to submit to lie detector tests because I like you have had enough.
[…] that they’ve replaced their CEO, replaced their marketing chief, and stopped the worst of the marketing nonsense I used to complain about. ParAccel has some interesting plans for ParAccel 3.0 which are, […]
[…] recently, I was extremely critical of ParAccel’s marketing. But there was an almost-clean sweep of the relevant ParAccel executives, and the specific worst […]