5x payout to investors? $900M for Tegile? Just throwing out wild numbers, eh Chris?
38 posts • joined 17 Jun 2015
OpenStack... from day zero, looking at their web site and such, my impression is that this thing has evolved into a vehicle to pull in supply side sponsorship to allow vendors to push product and/or services. Obviously there is a lot of supply-side interest to get on the bandwagon. AFAICT it remains to be seen whether there is actual, well, demand-side demand.
This theory can be proven wrong. Does anybody have numbers on, for example, how many VMs run under control of an OpenStack environment, vs competitors (VMware, Amazon, Microsoft, etc)?
I agree, this may actually be interesting. It seems to revolve around what constitutes a derived work. What is the same vs what is separate. Where is the dividing line exactly in the vmware product? Is there one? These are fundamental questions.
There is an old rms (I think) argument about why the LGPL is necessary: because any time you link code together in the same binary, a derivative work is created. I am not sure this view has any basis in law. My view has always been that this argument is intellectually vapid. Any good programmer knows how to create internal interfaces and keep parts of programs separate, even if they happy to run in the same address space and/or get linked into the same binary. What constitutes a derived work has to be answered with respect to an understanding of what *is* the original work, and, in combination with something else, is its identity preserved? What constitutes incidental mingling but fundamental separation, versus bona fide derivation?
a couple easy predictions
1) The smaller tier of on-premises suppliers will consolidate faster than predicted, as venture financing in this sector (high in the last 10y) dries up, and consolidation ensues.
2) The smaller and mid-tier of public cloud providers will grow in visibility. This is consistent with a number of comments above. More of this business is local than is widely perceived. Some of these providers will do better than others, and a few will go public.
3) Cisco will make an acquisition to get into data storage. It will not be NetApp.
check the math
A 5x payout on 1.5M is not a 7.5M acquisition. It's 7.5M divided by the fraction of the company that the investors in that 1.5M round got. Suppose they got a bit under 20%, so then it's about $40M assuming the 5x payout (which seems like a very rough estimate). In any case most of the valuation is captured in partially vested founder/early employee stock.
Re: Think data, not storage
Honest question: is moving your data once in a couple years really that big of a deal?
From where I sit, a lot of customers are forced to move portions of their data *much more frequently* due to issues of capacity, performance, availability (vendor bugs), correctness (vendor bugs), forklift upgrades (same vendor but cross-architectural boundaries), and more.
Nice piece Chris. Just to amplify the point, I would note that the magnitude of the installed base under threat is not just billions, but a number of *tens of billions* -- if you figure overall industry revenue has been approximately $20-$25B/year over the last few years, and multiply that by ~3y expected service life -- well, it's a pretty big number. It might not be exactly $75B because, frankly, due to competition and technology improvement, customers get more for their money these days. Call it $50B+.
Re: Er, that's actually a South Dakota/Minnesota billboard
Upper West? *What*? You sound like the kind of poseur who comes to San Francisco and starts referring to the place as "San Fran" or "Frisco".
Confusing the West with the Midwest is an extremely elementary error. I don't think "Upper West" is even a typo -- you went out of your way to capitalize "West". Midwest is one word, contains no hyphen, and there is no plausible typographical explanation for the capital "W". It is true that there is an Upper Midwest, although most residents therein (formerly including myself) would simply refer to themselves as being from the Midwest, or perhaps as being from the particular state from which they hail. Readers interested in the exact boundaries of these regions can find helpful entries on Wikipedia.
Re: 7 mode vs. C mode
This is an excellent question. The only theory I can come up with, and it's pretty tentative, is that that being an EMC customer appeals to people who expect this kind of thing from their storage vendor. They like being told how to do things. They like being dominated. They like taking it up the a** every now and then.
Re: it's not about the hardware
No, actually, it was a straight up question. Does it actually work in practice? Like it or not, you're conducting an experiment in software engineering. It's in your interest to convince the rest of the world that your experiment is a success. I'm mildly interested in the outcome (professional curiosity, you might say). Your fraction of the storage market is tiny and believe me, I have no reason to want you guys to fail.
Re: 92M in R&D for a single product vendor?
I'd be surprised. Whatever they have, whatever positive thing they can say about their business, is going to go into the S-1.
They are probably just not particularly efficient with their resources. Growing headcount super fast is a good way to spend a lot of money super fast. Every headcount they add requires incremental hardware resources (not cheap), lab space, HVAC overhead on that lab space, health insurance, etc. Basically, one of Pure's core competencies is spending investor's money. As somebody else said, that is exactly why they need to raise another round, hence the filing.
Yes, I work for a competitor. Which one? It doesn't matter.
This illustrates a fundamental weakness of hyperconverged offerings. Storage at high throughput tends not to be cheap computationally. This reflects the reality that storage vendors have to be focused on correctness and features, not just 100% iops. With CPUs being shared with application workloads, adding storage intensive applications ends up consuming CPU at double the rate -- once for the apps, and once for the storage.
Here's my take on Pernix. I work at a storage array vendor that supports VMware. As such we are basically required to implement a VAAI plugin (it runs on ESX hosts) in order to support array offload cloning of virtual machine disk files.
1) It is a total cluster fuck that this whole thing requires vendors to implement plugins. The VAAI ESX to array interface should have been defined as a network protocol. It would be extremely simple. Far, far simpler than NFS v3, and far, far, far, far simpler than VVOLs, which VMware is now trying to get vendors to implement.
2) The guy whose fault this is, I think, is the CTO of Pernix.
3) Occasionally we get a customer call where their ESX host is blowing up (purple screen) and *our* VAAI plugin was installed, among others, so they are calling everybody and looking for answers. However, reviewing the log files, we see messages that are obviously from a Pernix ESX plugin. These log messages contain the usual type of content. The message prefixes give file names and line numbers. In the messages we saw, the line numbers were IIRC in the 6000's range. Who the f puts 6000+ lines of code into one source file in one plugin? To boot, the code looked to be reimplementing TCP, at least going by the semantics of the message -- reporting things like sequence and ack numbers. So, we told the customer to talk to those guys, and haven't heard back.
4) Get a real architecture.
Some things never change.
1) Cisco doesn't know how to do storage.
2) The "Cisco acquisition machine" is vastly overrated.
3) Many acquisitions suck. Maybe the reason they're selling the company is because they know their product is weak and will not stand a chance in the market.
Thanks for the entertainment though!