You Call That Big Data?!?
Just to follow up, here's a fantastic blog from Nimble's VP of Support detailing the vast differences (and competitive stats) of InfoSight vs imitation.
8 posts • joined 31 May 2014
Disclaimer: I am an employee of Nimble Storage
Ok, so Pure is using 3D NAND TLC. I still fail to see why this qualifies as a big announcement, especially with the flare of such a production. Pure is by far not the only one, and certainly not the first, storage company to implement 3D NAND TLC into their product lineup. Kaminario has been using these drives for some time. Nimble Storage has already begun shipping them as well. So Pure is once again actually just playing catchup to other vendors, the only difference being a very expensive marketing production to make it look otherwise (some of us want to be fiscally responsible companies).
It is said that Imitation is the sincerest form of flattery, and we certainly thank Pure and others for acknowledging the disruption Nimble has brought to bare on the storage industry with InfoSight. However, comparing Pure1 to InfoSight as similar platforms is simply misleading. Case in point: Nimble collects 30,000,000 -1000,000,000 unique data points from each system on a daily basis (not counting per-VM analytics). Per Pure's own description of Pure1, they collect 10,000 data points. I'll let readers to the math to see if you can really compare the two platforms. There are a slew of other massive differences in the actual analytical capabilities of the two platforms that I will reserve for a different post if necessary, but suffice it to say that calling Pure1 an imitation of InfoSight is already giving it too much credit. As a side note, system upgrades are managed by Pure support because they have to run manual system and health checks before the upgrade. Why not build that into the platform itself and automate the entire process? Nimble has done that since day 1.
While SmartStack is a Nimble originated program, the point of differentiation is that it a JOINT agreement with Cisco. While not all of Cisco's teams may be out evangelizing SmartStack, you'll actually find some that do. I completely agree with all the sentiment that in all of these programs the storage vendor is the one that benefits most. My only point is there's a huge difference between something like SmartStack that, while initiated by Nimble, was completed and coordinated in conjunction WITH Cisco whereas this latest announcement is a one-sided marketing ploy. It bothers me not because there's another vendor working with Cisco (competition is good for the market) but because ultimately it misleads customers into believing there's something there that doesn't actually exist.
You're absolutely right in regards to the Dell + Nutanix deal. I was referring more to the wave of startups claiming they have a joint architecture WITH Cisco (as Pure's announcement would lead one to believe). There's a difference between something like FlexPod or SmartStack where Cisco is an actively participating partner offering joint support, validation, documentation, etc. and something like what was announced today which is a one sided engagement. Sticking someone's servers into your design does not make it a partnership. I would encourage people reach out to their Cisco contacts to get their take on this...as was eluded to earlier, there's a reason there's no announcement about this from Cisco in contrast to what they had done with other actual partners. Kudos for the marketing ploy, but shame for misleading the public.
There is a bit of a difference between many of these deals. How many are actually backed by Cisco? How many have joint support offerings through Cisco? How many have Cisco published documentation? One has to wonder why support has to go through channel for this.
HINT: Nimble has all of the above.
Disclaimer: I work for Nimble Storage, but my views are my own.
DISCLAIMER: I work for Nimble.
That's actually one thing (of many) that sets Nimble apart from a lot of the startups. The deals are being won not by dropping the price but for once by selling value. If you take a look at the margins reported in the investor reports you'll see what I mean. Profitability can easily be achieved by stopping investment into further development but the company has a much grander long term goal and hence has continued to put money back into the company/product/staff and has pushed profitability until 2016.
DISCLAIMER: Nimble Employee
I do urge you to read the full earnings report and analyst feedback. There's quite a bit more to this story if you don't focus on one factor. For example, margins continue to rise, are some of the highest seen in the industry, and continue to beat expectations. In other words, Nimble isn't selling just by dropping price...for once it's actually a company selling on value. If you look at the trend of various costs vs. revenue, you'll find they're all going in the exact direction that investors want (and ahead of expectations). Look at operating costs. Look at timing expectations for profitability. Keep in mind the age of the company and the fact that the IPO was less than 2 quarters ago. There's really more to it than just the revenue vs. loss rate in itself. But hey, everyone is entitled to their opinions.
DISCLAIMER: Nimble Storage Employee
Nimble's Fiscal Year is not the same as Calendar Year. FY15 ends on January 31st, 2015. In other words GA FC support is expected within the next 7 months at most.
While I don't disagree with you one bit that FC is a major enabler for the enterprise, I do think you'd be surprised at how many enterprise shops have been more than willing to take on iSCSI with Nimble. FC is going to vastly increase the reach for sure, you can clearly see the stats within the earnings report on enterprise adoption.
As for latency, I must vehemently disagree. If you are comparing iSCSI of yesteryear on 1Gb switches with low port buffers to 8Gb FC, then absolutely! If you have a properly implemented 10Gb iSCSI network optimized for storage traffic and compare to 8Gb FC, then your statement doesn't hold true. And at the end of the day, the few microseconds of difference you might be able to find are irrelevant compared to multi-millisecond latency in legacy storage.
As for AFAs, FC has nothing to do with performance improvements on a Nimble array. We can (and do) go head to head with them with iSCSI and win. FC is simply an enabler to go after environments that require FC.
Biting the hand that feeds IT © 1998–2021