NetApp to Acquire Bycast

Hot off the presses, NetApp to acquire Bycast. If you're unfamiliar with Bycast, they are a leading developer of object storage software designed to manage large scale global content repositories of images, video, and records for enterprises and service providers. This technology is currently available with their StorageGRID technology.

Object-based storage is an emerging approach to storing unstructured data. Unlike traditional NAS systems where on manages volumes or 'containers of storage'; object storage is a 'container-less' model that manages globally distributed data through a front end metadata database. This architecture allows applications to access data from the same location independent of where the data is actually stored, and offers much simpler data queries.

Policies can be used to control where data is stored, but the user doesn’t manage the data by containers. It also enables a globally-distributed data model by providing a location-independent access layer. Customers can have a consistent access method for their global users while they consolidate data centrally to drive efficiency and cost savings.

We have already been working with Bycast on a number of deals where the combination of our products offers a compelling solution. For now check out Bycast and the press release, and I'll share my thoughts on regarding the future storage with VMware later this week.

Comments

[Warren]

Sounds like NTAP can't scale on its own. Should make for an interesting combination of software.

[Vaughn Stewart]

@Warren - I dont quite understand where you see a scaling issue, maybe youd care to elaborate?

What Id suggest is theres a need to deploy storage clouds which are accessed by hypervisor and application clouds. Ill share more on this direction by Friday.

This message was sent by my thumbs... fault them for the brevity and poor spelling

[Warren]

I have friends at NetApp, so great news for them. I think this is a great play, coming from Objectivity where we sell an object oriented and distributed database we constantly here that ODB's days have come and gone. But the market is moving to the need to store petabytes of data that are mostly unstructured. And geographically dispersed. So products like Bycast are addressing the reality that all this data is in the form of objects, so storing them that way reduces complexity, increases performance etc. Doesn't mean NetApp couldn't do this before, but realistically the size of their customers data will be growing tremendously over the next 10 years (can you say exabyte?). So this merely positions them better for where they're going.

[Warren]

Just realized the first poster has the same name as me...funny - two different guys!

[Vaughn Stewart]

@Warren - well said, thank you.

[Jonas Irwin]

Interesting.
Netapp consistently slams of other "traditional array" vendors who sell (and own) more than one product to address different customer use cases. How does this product fit into the overall ntap one OS meets all use-cases strategy? That question aside, it certainly seems like interesting technology. I'm stoked to see you broaden your blog and talk about stuff other than snapshots (and all ntap technologies that hang off of snaps) and dedupe in the weeks to come. You seem like a good dude with lots of great insights. For your sake, I really hope this new product doesn't follow the same death spiral of other ntap acquisitions like Topio, Spinnaker, Alactritus etc.. (Please don't take this as criticism of you personally as it is not intended to be)

[Vaughn Stewart]

@Jonas

Hey man, hows things? No offense taken. Im stoked about the acquisition as Im confident that VMware will be datastore-less in the future. As for Spinnaker, its in ONTAP 8. I do look forward to ONTAP hooks being integrated into the Bycast technology. Im still stumped as to why EMC dogs dedupe as it reduces the storage footprint and network bandwidth requirements throughout the data center. I mean whats easier migrate a 25 GB VM or migrate the 12GBs which are unique to that VM?

Its nice to hear from you, even if its in this forum. Drop me a line if you shoot into RTP.

This message was sent by my thumbs... fault them for the brevity and poor spelling

[Jonas Irwin]

Things are great!
I stand corrected on Spinnaker.. Is that a separate code line than data ontap7? (Wait..you don't have to answer that ;-))

Actually from where I sit, EMC loves dedupe and compression, particularly in the backup environment. You've seen our primary storage rapid clone utility that exploits both dedupe, space saving pointer based (read/write) replicas and compression, right? Any emc'er who dogs dedupe, is simply misguided. Primary storage block based dedupe has it merits without doubt but its only part of the entire efficiency picture. I'm not suggesting that you don't realize this btw. If you're interested in object based stuff and other forms of storage platforms and optimization, I think you'd be pretty fired up by all the stuff going on over here.

If I do get out your way, I will definitely get your # from Chad!

[Vaughn Stewart]

Re: [NetApp - The Virtual Storage Guy] Jonas Irwin submitted a comment to NetApp to Acquire Bycast

@Jonas

Spinnaker has been available as our GX, which has a select set of customers most notably large compute grids like what one sees with movie special effects houses our scientific research groups, etc... The primary capabilities which we lead to the acquisition was for the Global Name Space and not-disruptive mobility of data. This is now available in Data ONTAP 8.0.

As for dedupe, you guys do kill me, and please dont take this personally. EMC pitches tech like dedupe, and then dont offer it on your flagship products like Symmetrix and Clariion. Spin Spin Spin goes the marketing machine. :)

Enjoy your weekend

[Chad Sakac]

@Vaughn, just so you can see how it sounds/feels from the other side, listen to your words with a slight twist:

"As for dedupe, you guys do kill me, and please dont take this personally. NetApp pitches tech like dedupe, and then dont offer it on your flagship products like Ontap 8 unless you put it in 7 Mode - negating features. Spin Spin Spin goes the marketing machine. :)"

ditto, but switch in "I'd like to use Data Motion and Dedupe".

These may sound like a nit to you, but:
- for every product inconsistency we have (and we at EMC have FAR too many, working hard to reduce them wherever we can without sacrificing use case priorities), I'm happy to point out one in return.
- for every time you rail at perceived inconsistency from us, I rail at it in the opposite direction.

(if only we could see the world through each other's eyes for a small period of time)

That, of course, isn't very productive (which is the root of our continous dialog with you on why I think NetApp needn't paint themselves in opposition to EMC and we shouldn't do it either).

I'm sure you're working on trying to make features between 7 mode and 8 mode consistent, but it's hard work and takes time.

Likewise, we're working on our side hard to make our features more consistent, but it's hard work, and takes time.

Capabilities appear first where it's a) technically easiest; b) the product prioritizations align (i.e. efficiency means different things in different use cases).

Within EMC production storage platforms, you can count on:
- Automated sub-file tiering
- Large system-wide read/write cache models
- Production Compression
- Virtual Provisioning and zero reclaim
- Deduplication (today at the file level, but we're not stopping there)
- large scale snapshots
- Dense storage configurations
- Non-disruptive mobility (something we do within platforms today, across platforms tomorrow)

...to become increasingly universal as core storage services. This is rooted in the fact that internally, core engineering work has been underway for years to share foundational elements (both hardware and software) wherever possible. For example, Celerra and CLARiiON share foundation block-layout/filesystem logic in the virtual provisioning layer of their stacks. This layer of code is used for a broad set of capabilities, including many of the above. There are cases where the use cases ARE aligned enough where more commonality is good.

Over 2010, more and more of this will become visible.

BUT, the Bycast buy (and I think it was a smart move on your part) highlights that sometimes ONE core technology stack isn't enough (or isn't even right for the task). EMC Atmos, is likewise a different technology stack - that can leverage any storage (ranging from JBOD to RAID, to standard kit, to enterprise arrays). How long Atmos and Bycast run on standalone servers is surely not forever.

Another example in even the adjacent production storage use case served by enterprise storage arrays are different than the midrange. There you have requirements like:

- serving IO via all ports, all storage processors symmetrically (implying shared global cache)
- no element failure affecting the overall envelope of the total system
- host device counts in the tens of thousands
- local and remote replicas in the tens of thousands
- and yes, host attach across a broader set of hosts.

Those are, in that use, case "efficicency technologies", just in a different dimension than in the mid-range.

Another example is the case where we fought furiously for Data Domain.

I'm sure that had that ended differently, you wouldn't be saying that "consistent management, single platform, consistent featureset" is the most important thing - but like Bycast and like in the high-end porduction use cases there's value in "purpose built".

Fact is that in the backup target use case, priorities are different than production platform priorities - they conflict, and those differences are tending to accelerate.

For example in that use case, it's all about inline dedupe rates/ingest rates, scale of dedupe, and dedupe replication (lots of fan in use cases).

That's not to say you can't build a good backup solution out of a general purpose platform, and that both EMC and NetApp don't have great efficiency technologies that can be applied to that use case.

BUT, if that were always the best way - why would we have fought so furiously for that one? It's not rational (and we're both rational) to say that it was "to keep it from the other guy".

Look - my point (long winded as usual) here is that if you think Marketing (including describing features that are mutually exclusive, but neglecting to mention those exclusions) doesn't **also** happen at NetApp, or that innovation doesn't **also** happen at your competitors, well... you just keep on thinking that. NetApp is a strong competitor, and I'm happy to see you underestimate the competitive landscape for all of us.

Enjoy your weekend.

[Vaughn Stewart]

@Chad - As always, thanks for chiming in.

Thank you for the inquiry, as you know I believe customers who are going to spend ten of thousand to millions of dollars on a storage array deserve to know exactly what their storage technologies can and cannot do.

So why would you disseminate FUD when it is clearly stated in a publicly available document that Data Deduplication is supported with Data Motion. See the page 18 in TR-3814.

As far as Data ONTAP 8, 7-Mode and Cluster-Mode... yes there as differences in the 8.0 release however; I would ask you to cite any documents which my team has created which discusses deploying VMware on Cluster-Mode. This is very different than the message you communicate spin when you say "...for VMware we offer dedupe on our arrays..." Would you care to comment as to why you message in such a manner?

This is the same issue I have with EMC with a number of the authoritative messages around best practices ranging from SAN/NAS, dedupe, multipathing software, RAID levels, etc... where these best practices do not exist in public documents.

Customers considering NetApp and EMC for their VMware footprints understand what the are receiving from NetApp and how it will work. I'm not sure the same can be said with EMC solutions.

Wrapping up this comment, I'd like you to stop hiding behind the "..why go negative.." statements and claims. There is nothing negative in comparing two items in order to understand their similarities and differences. Everyone does it in every day. Do I take the freeway or the highway, do I have the chicken or the fish, etc...

I fear you are losing credibility with such tactics. Be transparent and let the technology speak for itself.