Comment: It's the end of one-size-fits-all WAFL waffle: NetApp HQ at Sunnyvale enters a new era today, in which it sells diversified storage systems. WAFL is no longer the defining factor.
WAFL (Write Anywhere File Layout) has been synonymous with NetApp's success as progressively more and more functionality has been added onto it. NetApp's FAS arrays run ONTAP, the NetApp O/S which has WAFL at its heart, and NetApp's ONTAP arrays have been positioned by NetApp as "the" unified storage solution.
You need filer access? ONTAP. iSCSI SAN? ONTAP. Fibre Channel SAN? ONTAP. RAID 6? ONTAP. Snapshots, replication, cloning, deduplication, flash caching? ONTAP. Storing files for virtualised servers? ONTAP.
ONTAP software expanded sideways across the storage stack and upwards as well. One size fits all under a unified software functionality and management scheme, for which NetApp charged big bucks, and for which customers were happy to pay – witness NetApp's recent string of superlative quarterly results.
The one fly in this ointment was that it took an age for external technology to be integrated, with the most outrageous example being the Spinnaker clustering scheme. Even now, years after the technology was acquired, there are still two versions of ONTAP, and NetApp's scale-out file storage capabilities are lacking in comparison to EMC's Isilon, DataDirect Networks, and HP's Ibrix.
It is even rumoured that there are fewer than a hundred actual ONTAP GX clusters in existence.
Unified becomes diversified
NetApp recently bought object and grid storage company ByCast and observers settled themselves down for another painful integration saga – or would NetApp not integrate the ByCast technology and run it as a separate product line?
Now here it has bought Engenio with its distinctly different storage array technology from NetApp's FAS boxes. Doesn't this utterly destroy NetApp's unified storage message?
NetApp co-founder and EVP Dave Hitz directly addressed this issue in his blog and says he doesn't think NetApp will suffer from the same issues as EMC "with its fragmented product line". That is because EMC has overlapping and differently functioning and managed features in its product lines, Hitz says, like multiple kinds of deduplication.
In NetApp, ONTAP has all such features while Engenio's arrays are like stripped down, high performance storage engines – very simple ones with few features. Feature overlap is "not an issue for NetApp. If you want the advanced data management features, go with ONTAP which has them all," says Hitz.
He admits he's over-simplifying, but says "NetApp and Engenio have very different design points optimised for different workloads and market segments." For customers that need big bandwidth without fancy features, Engenio is perfect.
Over time he expects ONTAP to be used more as a virtual machine (VM); it already exists as a VM for Fujitsu use, and the Engenio product could be a great match for that.
In other words, what was NetApp's universal unified storage message is being confined to the ONTAP FAS world and is unaffected by Engenio. The once universal unified storage message is no more... with Hitz now asserting that it doesn't matter.
One WAFL size does not fit all
We could take a view that Warmenhoven-era WAFL dogma has held NetApp back, and it is only since Tom Georgens replaced Dan Warmenhoven as the CEO that NetApp has been able to break its WAFL shackles.
Georgens has now admitted ONTAP FAS arrays could not serve customers well in the high-performance computing (HPC), full motion video capture, and video surveillance markets. Storage arrays sold into them need to have higher performance and more bandwidth than NetApp's FAS arrays; this is an admission of weakness by NetApp.
Georgens thinks being able to sell into these markets will increase NetApp's total addressable market (TAM) by $5bn by 2014. That's not a lot.
Let's take a fresh look at NetApp's product line and the storage market in the light of this, and see what other gaps might exist, now that Georgens has lifted the WAFL veil.
First off is scale-out NAS, where NetApp does not have, in El Reg's view, a product in the Isilon/Ibrix class. Engenio arrays could provide the hardware for this but new software is needed, unless the relative cluster turkey that is ONTAP GX turns into a swan in front of our eyes.
Secondly, NetApp still does not have any high end, multi-controller block access storage like EMC's VMAX, HDS' VSP, or IBM's DS8000. Engenio makes traditional dual controller arrays in the VNX/EVA class and there is no indication that a multi-controller engine system is coming.
Thirdly, NetApp lacks truly effective primary deduplication. Talk to Permabit guys; it's the only game left in town and could leave A-SIS for dust.
Fourthly, NetApp lacks any array federation capability. Also, it lacks a stack-em-high, sell-em-cheap, scale-em-up-and-out cloud-storage building-block box, in the Atmos mould. Lastly, it lacks a cheap and scalable iSCSI storage box in the Dell EqualLogic/HP LeftHand mould. Again the Engenio hardware could be a base for this but new software is needed.
NetApp just signalled it wants to be a diversified storage product line player. That's going to cause a sea change in the way customers, commentators and analysts view the company, because they will compare it to other diversified product line players, such as Dell, EMC, HP, and IBM, instead of being forced to compare NetApp to ... well, NetApp, because there was no other comparable company.
NetApp is late to the diversified storage party and the removal of WAFL waffle from its marketing messages is going to expose it to a harsher world. If it is going to compete in that world it has to have a strategy to decide which segments to go after and how to execute its business in those segments. It also has to work out how its customers can manage multiple NetApp storage product lines.
Somehow I think more acquisitions are coming as NetApp finds it necessary to bulk out its offerings faster than it can build them in-house. Welcome to the party. ®