EMC: The Storage Most Integrated with VMware?

Oh Vaughn – this is why I try my best to leave the “compare” stuff to others. It was also the key thing I was trying to convey when we talked live after the last blog post.
You and I are above this. We can’t stop a rep out there from making claims they shouldn’t.
Smart customers should listen to the technical people on the solutions, and work with the sales people on pricing and making the business case.
“Ignore sales claims” is a general best practice for me.
Why is this silly and beneath you (if you ask me):
1 – There are examples on your chart that don’t apply evenly to NetApp, either. For example (please correct me if I’m wrong) – VM-level flexclone operations via RCU are dependent on NFS – FC and iSCSI are datastore only. Ditto with “IO offload for VM clones” – currently (and this will be for a long time, as you and I both know based on VMware’s roadmap), you do this via file-level snapshots (file-level flexclones), not the future VMware VAAI stuff for fast/full copy (which is block centric). In some cases you don’t note that NetApp hasn’t done a given function (SRM failback, for example)
2 – you don’t even mention NetApp SMVI or EMC Replication Manager (which have some overlap with RCU and our VDI tool). This is a “check” for “datastore level snapshot/clone with ESX integration” across all our platforms, and protocols (including NFS).
3 – Does vSphere ESX auto register initiator records and group host entries with NetApp?
4 – Has NetApp integrated Filerview with the vCenter APIs (or, since you started this, updated it’s look/feel in the last few years for that matter)? We have in Navisphere and Recoverpoint, and are bringing that out across the board.
5 – We have both demonstrated the VAAI integration for offload using Write Same/Zero, Full/Fast Copy. I’ll be aiming to add ATS and TP stun at VMworld this year, and I’m sure you’ll do the same. That’s a couple of missed checks on your list.
6 – the fact that we charge for PP/VE doesn’t eliminate the fact that it is an integration that currently, NetApp has chosen not to make. Other have done something similar (Dell/EqualLogic) and seem to give it away for free. In the end, the market decides value – but I’m not sure it makes sense to say “their integration doesn’t count because it’s not free”.
Above and beyond that EMC integrates with the vStorage APIs for Data Protection in Avamar and Networker and NetApp does not – because you are a storage vendor, not a traditional backup vendor (partner with Symantec)
Another one? We have vCenter plugins for VM configuration control and compliance, application correlation and more. Where do those fit? Well – they kinda don’t – EMC is more than a storage pure-play (though of course we’re VERY focused on that part of our business). NetApp is getting started with the Onaro acquisition, which is very good for you strategically. storage pure-play is not a good long term strategic position.
Most importantly the reason for you and I to stay out of this fray – it’s SO easy to be wrong when you focus on the other guy: a) because the details are easy to miss (we do a view manager import for example – along with the stuff I noted above), and b) because it changes SO fast – major updates to EMC VDI deployment plugin – our parallel to the RCU, and the EMC Storage Viewer – our equivalent to the NetApp VSC are right around the corner, along with many other things.
These **not coincidentally** eliminate many of the items in your checklist. One area that we debated, but ultimately elected NOT to add to the EMC storage viewer plugin (the VSC analog) is provisioning tasks – for most of our customers, the feedback was that while it was neat to demo, the it did cross a line of accountability that they didn’t like. They wanted VMware folks to be informed on storage, and Storage people to be informed on VMware to make life easier, but that the ability to reconfigure or take action in either direction was difficult to accept. This will change over time, but that was a conscious decision on our part.
Now – why isn’t it a coincidence that the updates (VMworld really is jut around the corner, isn’t it 🙂 eliminate a lot on your chart?
It’s not a coincidence because smart competitors look that the good things others do. Where we are first – great. Where other come before us – and customers dig it – we will do the same, and try to 1-up.
If you think that NetApp doesn’t do that I think you’re kidding yourself.
If NetApp REALLY doesn’t do that – it’s VERY dangerous. If any company thinks they have a monopoly on innovation – that they are permanently, intrinsically superior to everyone (EMC and everyone else) that’s the arrogance that can kill you (it did kill EMC almost once long ago, and we learned a lot from the experience).
Take my advice – this is now 3 recent posts which are thematically structured around “EMC sucks, NetApp rocks”.
You can point out how NetApp rocks without the “EMC sucks” – it doesn’t help your case.
I’ve said it before, and I’ll say it again – NetApp is a great company, has great products, and is a great competitor. You don’t need to stoop to these sort of things to talk about what makes your products great.
Don’t tell me about what OTHERS (competitive team folks, reps in the field, whatever) do – I’m talking about you and me now.

Although I’d much rather see a technical in-depth topic being discussed I must say this is also useful to get the discussion going and getting a real life comparison especially if you add Chad’s comments to the mix.
For most of my customers this isn’t an issue at all. Most of them care about reliability, performance and usability in terms of management. Most storage admins couldn’t care less about a virtual environment unfortunately. They care about their storage and most of them don’t even want to integrate with the virtual environment as that takes away the traditional separation between the teams. (Seeing the same with networking, a lot of them are reluctant to use the Cisco Nexus)
Although I think the current level of integration of both EMC and NetApp is a big step forward both are not mature yet.

To a certain extent I have to agree with Duncan–despite the best efforts of a lot of integrators, the storage admins, network admins, and the server admins are still very siloed. Bringing a lot of storage mgmt into vCenter Server is helpful at the lower end of the market, but not so helpful at the higher end of the market. And after years of telling the network admins to go away and the server admins to handle the networking, many orgs are now unwilling to undo that effort by bringing the Nexus 1000V in-house.
I’ll further echo the sentiment that all vendors should focus on their own strengths rather than focusing on the weaknesses of their competitors. I’ve told VMware the same thing repeatedly. Let your products stand or fall on their own merits.

Chad, Duncan, & Scott Thanks for the feedback. I’d like to touch a few of the key points…
Chad – This post is about full disclosure. While your blog posts are pretty clear, I’m not sold that our customers are receiving the same level of distinction in their meetings. I’ve always said EMC makes solid products. Inferring a sentiment of EMC sucks t these posts is not an accurate depiction of the content.
Duncan – I’m with you, tech post are much more rewarding to write and read
Scott – While spaning IT demarcation points is a challenge we are hearing a loud demand for process automation. I agree we are hearing this the loudest from small shops and early adopters of cloud based architectures. I expect those who are late adopters to adopt later.
I truly respect and appreciate your feedback, so thank you.

Scott &/or Duncan,
While it’s normal to expect people to resist change, how long to do expect companies to remain silo’d in their operational models? For things like VoIP (integrate voice & networking groups), it took about 18-24 months. I could see this being roughly the same timeframe, although external cloud services (ie. hosted storage or hosted computing) could accelerate that if some departments feel like they can be more competitive with an external service.
Thoughts?

Chad,
You hit the nail on the head, this is just another example of 1-upmanship. Although I am mystified why you don’t see this as anything but a competitor counter claim…
Other competitors will be doing it too, you better get used to it 🙂
This ‘VMware works better on our own kit’ message is certainly hitting the street, but I think you’re comment to ignore the (rogue?) sales rep is a little naive, it lines up with the corporate messaging: http://www.itnews.com.au/News/146549,inform-emc-plays-up-ties-to-vmware-cisco.aspx
(Not a NetApp employee)

@Vaughn, @Paul P – I understand the need for vendors to build these sort of tables. That wasn’t my point.
The point was that I think Vaughn is above this.
The problem with all these sorts of tables, is they:
1) try to compare things that intrinsically are VERY different – for example, architecturally a Symmetrix is VERY different from a Celerra or a NetApp array. I’m not claiming better/worse – but that some things that are easy on one, are not easy on another.
2) Every table I’ve seen (this one included – but I’ll lump in the ones our competitive teams produce) select the rows to suit purpose. My earlier comment pointed out a whole slew of rows that were neglected (element managers integrated with the vCenter APIs for example), and there is no NetApp column, so the presumption is they are a “check across the board” which they are not.
3) They are out of date the moment they are produced (I noted several in the earlier comment).
4) They miss huge chunks of color. If you look at Scott and Duncan’s comments – they highlight WHY we didn’t make “action-based provisioning for storage” in our vCenter plugins (and conversely “action-based provisioning for VMware” in our array element managers that integrate with vCenter APIs) universal. To DATE customer input has been universal. Some things are “ok” (like snapshots, VM-level dedupe) from the other context (in this case the VMware Admin context), and others are not (like extend/shrink filesystem/LUN, provision new storage object). I AGREE with Vaughn (and this is a major focus at the engineering level, and also will be sharing at VMworld) that over time, infrastructure needs to become more automated, shifting input up to the VMware and Application administrators. but – today, we’ve gotten a lot of cutomer push-back when we try to drive that agenda too fast.
The main thing I’ve been talking a LOT about with Vaughn (publicly and privately) is that NetApp seems to define themselves as “anti-EMC” (and more broadly “anti-everyone else”. I don’t think that helps them. They don’t need to define themselves AGAINST others – they can stand on their own.
To highlight (and I hate doing this, because it goes against what I’m trying to accomplish here), note that when I publicly talked about our element managers (the actual platform management interfaces) are integrate with vCenter APIs (on TOP of the vCenter plugins) here:
http://virtualgeek.typepad.com/virtual_geek/2009/07/state-of-the-art-integrating-storage-and-vmware-part-i.html
There was NO statement like “hey, we do this, and NetApp doesn’t – HA!”. In fact, I didn’t say “hey, we do this and NO ONE does it – HA!”. Both statements are true (at least for today – though I’m sure over time that will change).
Vaughn’s last couple of paragraphs: “I am fairly certain they no longer believe that EMC is the storage most integrated with VMware”. So – is the claim that NetApp is the most/more integrated? If so – what has NetApp got along the lines of the places I’ve highlighted where there is no NetApp analog?
Also, he suggests that he is the one explaining EMC technologies to the customer. Boy, I would be terrified to explain why NetApp would not be good to a customer – as a) they are fine of course; b) I would be likely to be wrong, out of date, or be missing important facts. On numerous occasions, where I’ve seen incorrect FUD eminating from us, I’ve reached across the aisle to make sure that we’re correct, but if THAT’S what’s happening in campaigns (not “let me tell you why customer’s choose _____”, but rather “let me tell you what’s wrong with _____”), that’s not good for NetApp (or EMC when we do it.
I think that taking that approach only lessens the debate, and lowers the standard of dialog.
Look – I’m not naive:
– I know that people need/want these little “fit it on a page comparison tables”.
– I know that sometimes, both in the past, and even today, EMC is the one casting a stone incorrectly – I’m not claiming “holier than thou” on a company basis.
– Sometimes, I fall down, and cross the line I’m pointing out here. I’m not perfect – we’re all sinners 🙂 But, I’ve made it a conscious effort to always read what I write and say “what would I say, reading this, if I’m not an EMC person?”. That’s a challenge I’ve put out to Vaughn.
My point was that I think that it was beneath Vaughn as an individual, and that this tendency to constantly try to make everything into neat tables, with the emphasis of denegrating others by any vendor – is bad. Not just because of outcome (where inevitably one vendor comes out on top in some columns, and the others in other columns), but because as a general principle it doesn’t advance the greater good, or serve the customer in any useful way.
Not sure if I’m helping, but trying to explain my rationale.
Vaughn, at VMworld, print out your sheet. I will show you the errors, I will suggest other rows I think you’re missing, and I will show you what’s right around the corner (so close that it should be included in a table) – that’s IF you want to continue the sillyness.

Chad – Do you feel your post on Scott Lowe’s blog is sharing all of the info your cusotmers need to make a solid decision?
———————————————
From: http://blog.scottlowe.org/2009/08/17/netapp-rcu-and-vsc
– at present my comments have yet to be approved by Scott so I am duplicated them here –
———————————————
Chad Sakac on Monday, August 17, 2009 at 7:48 pm
These are very good utilities. I would be remiss (as an EMCer) if I didn’t point people interested in this idea, but are EMC customers that there are several analagous vCenter plugins for EMC.
The EMC analogue to the VSC is a vCenter plugin called the EMC Storage Viewer launched at VMworld Europe 2009 in Feb.. It is freely available on Powerlink (details are on my blog here: http://virtualgeek.typepad.com/virtual_geek/2009/04/where-to-get-the-emc-storage-viewer-vcenter-plugin.html)
The EMC analogue to RCU is a vCenter plugin called the VDI Deployment Tool launched at VMworld Europe 2009 in Feb.. It is freely available on Powerlink (details are on my blog here: – http://virtualgeek.typepad.com/virtual_geek/2009/03/vmworld-europe-2009-emc-post-show-report.html
The EMC Analogue to SMVI (snapmanager for Virtual Infrastructure) is EMC Replication Manager.
Personally, I think the RCU is a bit better than our existing deployment tool (VMworld is coming up 🙂 but on the others, I think rational folks could debate back and forth which are better.
(we have a couple others as well at EMC with no NetApp analog – the SRM failback vcenter plugin most notably – with a few others coming. Like VSC, we plan to bring these plugins together over time via a common framework model).
It’s fun to have a strong competitor to parry back and forth with – in the end, it’s good for customers!
———————————————
Vaughn on Wednesday, August 19, 2009 at 10:29 am
Sorry to be late to the party – a few points and thoughts
1. EMC products often only work on singular arrays and/or protocols.
The EMC Storage Viewer is similar to the NetApp VSC with a few ‘minor’ differences including… the VSC works on every platform running Data ONTAP, with every storage protocol, and actually changes the configurations of the ESX/ESXi hosts to their optimal values.
The EMC VDI tool is similar to the NetApp RCU with a few ‘minor’ differences including… the EMC tool clones 2 VMs per LUN, only supports Celerra, and iSCSI. The NetApp RCU works on every platform running Data ONTAP, with every storage protocol, can clone any number of VM and datastores, imports directly into View Manager and XenDesktop, and all clones both Virtual Servers along with Virtual Desktops are pre-deduplicated.
Is the EMC Replication Manager similar to SMVI? I don’t know. Can it use its off-site backup disk with SRM? Does testing a SRM recovery plan require additional storage or disrupt replication?
2. The VSC & RCU are both free and SMVI is aggressively priced!
3. The EMC failback plug-in does automate the return of services to the primary site. Come check out the NetApp booth at VMworld for more on this type of functionality! Sorry – I can’t disclose everything here!
———————————————
Chad – Is information sharing whcih compares one offering to another still classified as unprofessional by EMC standards?

Answers inline. I always try to be as transparent as I can. Will you?
** I really want to state something up front. I DO NOT WANT TO DO THIS. This is being negative on a respected competitor, but you have consistently poked and goaded. my STRONG preference is for the customer to challenge each vendor they consider to make their TOTAL argument why they are the right choice. **
Sigh – here it goes.
VAUGHN: 1. EMC products often only work on singular arrays and/or protocols.
CHAD: True. Our platforms have relative strengths and weaknesses in various use cases. Some customer need platforms for all sorts of use cases. Example: constant non-disruptive operations, very high active port scaling, huge numbers of storage objects, replicas, and mainframe support – NO ONE does this in midrange platforms). NetApp calls this a weakness always. I think a more rational person looks at it and says: “sometimes it’s a strength, sometimes it’s a weakness. Depends on the customer requirements”.
What is rarely discussed by NetApp – is not all functionality is common across protocols. Example – VM-level replicas is NFS only – not iSCSI/FC/FCoE. Another Example? MultiStore (is that the right name?) is iSCSI/NFS only.
VAUGHN: The EMC Storage Viewer is similar to the NetApp VSC with a few ‘minor’ differences including… the VSC works on every platform running Data ONTAP, with every storage protocol, and actually changes the configurations of the ESX/ESXi hosts to their optimal values.
CHAD: EMC Storage Viewer works on EMC CLARiiON, Symmetrix, and Celerra – every EMC platform. There is currently MORE visibility in CLARiiON and Symm than there is with NFS datastores, but the correlation of objects is in there.
customer feedback (at least to date) has been that they don’t want the VMware administrator provisioning storage with the exception of triggering snapshot events and triggering VM failover. We’re pushing that, but getting pushback from the customers.
VAUGHN: The EMC VDI tool is similar to the NetApp RCU with a few ‘minor’ differences including… the EMC tool clones 2 VMs per LUN, only supports Celerra, and iSCSI. The NetApp RCU works on every platform running Data ONTAP, with every storage protocol, can clone any number of VM and datastores, imports directly into View Manager and XenDesktop, and all clones both Virtual Servers along with Virtual Desktops are pre-deduplicated.
CHAD: First of all, the current VDI tool has no restriction on the number of VMs per datastore. It currently is iSCSI and Celerra only. RCU 2.1 is coming at VMworld, right? The next version of our tool (which is being renamed since it’s not just about VDI) enables: NFS and iSCSI (still Celerra only), VM level offloaded snapshot/clone, datastore level replicas, and some (we’re pushing the same customer boundary here) provisioning actions (similar to the RCU 2.1 functions you discussed). Based on feedback, we may be broadening the function.
VAUGHN: Is the EMC Replication Manager similar to SMVI? I don’t know. Can it use its off-site backup disk with SRM? Does testing a SRM recovery plan require additional storage or disrupt replication?
CHAD: I think they are similar. Some of your input is incorrect, some is correct. Can you use off-site replica with SRM? No. Testing an SRM recovery plan does not disrupt replication or require additional storage. So – here’s the ball back to you. Does SMVI handle Exchange? SQL Server? Oracle? (all as VMs if you want?) RM does. Can NetApp do this? Sure, with SME, SMO, SMS, etc. Ironically, many tools, for a single purpose.
VAUGHN: 2. The VSC & RCU are both free and SMVI is aggressively priced!
CHAD: EMC Storage Viewer, VDI are both free, and RM is considerably less expensive (in my experience) than SMVI, PARTICULARLY when you need SME, SMO, SMS, SMVI, etc.
VAUGHN: 3. The EMC failback plug-in does automate the return of services to the primary site. Come check out the NetApp booth at VMworld
CHAD: our intent with the tool was not to replace SRM, and you know as I do what the SRM roadmap is. BTW – don’t assume that our SRM failback tool is static, and we won’t be showing OUR update (including multiple platforms). But hey, since we’re talking about NOW (you talk about your futures, we can talk about ours too).
Come check out the EMC booth at VMworld for more on this type of functionality! Sorry – I can’t disclose everything here!

Hi Chad,
Could you share how Replication Manager provides application consistency? Does it rely on VSS within the guest or does it have its own mechanism?
Thanks!

John – app consistency is done in ways (and just one more reason why I think they are similar) that are similar to the SnapManager family (warning – I don’t claim to be an expert on SnapManager – and NetApp folks who know more – please correct me)
Exchange – uses VSS (and includes Ontrack PowerControls for mailbox restore – just like SME with the mailbox recovery option – which is the same Ontrack software).
SQL Server – uses VDI (SQL Server supports VSS and VDI, VDI remains the preferred method as it covers some scenarios VSS does not)
Oracle – uses the Hot-Backup mechanism.
VMware – integrates with the ESX snapshot mechanism in the pre-post array replica mechanism (for IO consistency), which in turn uses the VMware tools (which actually use a bit of EMC code) for VSS handling inside the guest. This guest-level consistency is INDEPENDENT of the application-level consistency mechanisms listed below. For very VERY dense datastores (many VMs) this mechanism can be turned off (exchanging faster jobs for lack of guaranteed VM-level consistency). I believe SMVI does something similar. BTW – not related to Replication Manager, our upcoming update of the vCenter plugin I referred to in this thread that does this at VM-level on NFS datastores only – just like RCU – uses a similar mechanism for VM-level consistency. You can see this at VMworld.
Sharepoint – uses VSS for correlation across the sharepoint farm, VDI for the SQL Server elements (and includes Ontrack PowerControls for individual item restore – I believe the SnapManager equivalent also OEMs a 3rd party for single item restore).
That last one (Sharepoint) is in the latest version that also adds NFS datastore support in the VMware use case (previous versions were VMFS only).
The funny thing is I work hard to keep pretty plugged into things here at EMC, and even I didn’t know about the Sharepoint piece until it crossed my path in Outlook today. This highlights the hazard of these negative FUD campaigns. it’s SO easy to be wrong, so easy to be right – but only in the most transient fashion as these move SO fast.
That’s my original intent with this comment thread. There’s so much goodness on use-case integration from NetApp, so much from EMC, and it’s moving SO fast – it’s hard to even keep up with your OWN stuff. People shouldn’t listen to what EMC people say about others – and the reverse is also true in the opposite direction with NetApp.
Customers should push vendors to demonstrate how their solutions work for THEM.

Hi there.
I’ve done a little review of the RCU, and video too…
http://www.rtfm-ed.co.uk/?p=1605

I believe the AMD Virtual Experience format is an innovative way to educate a new generation of buyers looking for a convenient online alternative to the traditional tradeshow experience.