Watch out EMC and vendor pals, Cisco's set to flog you with Whiptail El Reg's storage desk is hearing Cisco people saying their partnerships with storage vendors are stronger than ever, following the Whiptail acquisition. They say EMC and other storage array suppliers understand that Whiptail is a UCS-booster, not an array. Well, excuse me, but what? The idea of Whiptail as a Unified Computing …
My guess is that Cisco was using a bit of spin to say that they aren't competing with the core offerings of EMC, etc. as Cisco defines them (i.e. big arrays of drives for long term storage a few layers down). However, all the players are moving towards the emerging market of "fast data." Given the Nicira acquisition, I don't think there's much love lost between the likes of EMC and Cisco.
The Whiptail acquisition was never about networked storage or to enter the SAN/NAS space (Whiptail itself was never really good in that area). Also the way that Whiptail scaled (using Infiniband) is obviously going to be culled as Cisco's world is FCoE/Ethernet.
The biggest threat to the Whiptail purchase is aimed primarily at FusionIO; the biggest margin killer in a Cisco UCS deal and a major drain in PCI mezzanine slots within UCS blades. The common prediction rumoured is that Whiptail will be merged into a FusionIO-esque server device to accelerate required applications.
Wow. You people don't think laterally. You turn Whiptail into a "like a FusionIO card" by making it one blade per chassis. That will then connect via the chassis' internal network infrastructure to the other blades in the chassis. The chassis-local Whiptail accelerates the local UCS nodes while also going using cross-chassis communication channels to make sure data is replicated in another blade in one of the other chassis.
This isn't fucking rocket science here.
It's also how you make the claim that "Whiptail isn't there to compete with EMC." It's not...at least not EMC's spinning disk stuff. EMC will provide you bulk storage that all your chassis will talk to. Whiptail will provide chassis-local acceleration and serve as a new tier of "fast storage".
Whiptail isn't a broadside at EMC, it's slitting Pernix Data's throat, along with it the margins that currently go to FusionIO, Micron or LSI in the form of their PCI-E cards.
That EMC wants to get into the same space is of no consequence. there are plenty of players.
Mark my words; Whiptail will evolve from a SAN into something closer to an acceleration tier. A hybrid, if you will, between Pernix and Tintri. Dangerous stuff...
I get that it could become a storage blade, but you still need a method of accessing the blade, so likely you didn't save a PCI-E slot or maybe you slide off a vNic and go iSCSI. But you've also just added much more latency by having to traverse the iSCSI stack and then the chassis and if you then have to mirror to a second blade for H/A that's even more latency vs FusionIO which is on the bus. Will it be "good enough" for a large percentage of requirements, probably.
I still can't see Cisco just throwing away Whiptail's current products in favour of a pure accelerator product, they could easily roll out the current family on Cisco servers. So all this "honest Guv we don't want the storage business" is smoke and mirrors. The irony being both EMC and Netapp bought the story and introduced Cisco (outside of their traditional networking role) into their own accounts.
Network, compute, storage. The Trinity of IT Infrastructure. Cisco has put a complete data center product together that can exist, standalone, to service a high performance data center and provide a monolithic vendor data center solution.
Solid State Disks are getting bigger and cheaper faster than EMC (or NetApp) seemed to anticipate. Several startups (including Whiptail) have developed product that aggregates SSD to directly compete with the traditional vendors. The traditional spinning disk vendors are losing business in cases where IO rates are critical - and, if you're spending a million quid on storage, IO rates ARE important. Massive stacks of spinning disks are being replaced with someone's managed stacks of solid state storage.
Most spinning disks will spin into oblivion as SSD's get larger, less expensive and more robust. Cisco is perfectly positioned to ignore the current technology in enterprise disk arrays and be a leader in the Enterprise Plus data center. Yep, sell your EMC stock until they integrate XtremIO technology into a large scale product.
Now, what;'s going happen to the Cisco content of NetApp's FlexPod?
put this together with Seagate & the Kinetic stuff for volume / slower storage and things become even more interesting. Seagate's market cap at $17.77Bn makes a slightly bigger task though ...
Although you'd think a company with that sort of market cap would at least do the decent thing and not run the Kinetic wiki on an eval licence:
© 2013 Seagate Technology LLC. All right reserved.
EVALUATION LICENSE Are you enjoying Confluence? Please consider purchasing it today.
Powered by Atlassian Confluence 5.2.5, Team Collaboration Software Report a bug Legal and Privacy
Each of you have very valid points, BUT lets look at the landscape.
1. VMware buys Nicira for software defined networking (July 2012....EMC owns 80% of VMware)
2. EMC forges server partnership with Lenovo (August 2012)
3. EMC's history of 'breaking-up'...late 90's with HP, mid 2000's with Dell, Lenovo partnership inked with EMC...seems like both companies see the writing in the sand.
4. NetApp....well they are just coasting....nothing really interesting about them.
Cisco owns the market when it comes to networking. However, companies like Arista Networks, Brocade, and obviously the legacy players cut into the market share. So how can Cisco continue to grow? Answer: One stop shop for the whole stack and take the margin on storage.
Why Cisco would buy Whiptail for $415 Million is beyond me. Whiptail has a fast and cool product but the amount of customers they have does not warrant that type of money.
Prediction: Cisco will end up buying one of the newer hybrid storage companies. Looking at the storage landscape, I suspect it will be Tegile. They aren't too small or big and support all backend connectivity. Nimble was my initial thought but that changed when they filed for a $150Mil IPO.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
