EMC vs Compellent

[Brovane]

Wow... That's some serious FUD.

Maybe that was the happy EMC customer.

 

[RobertR1]

If we're going to look at 5th and 6th year, probably pretty wise to consider all aspects. As Dell falls further and further, and cuts R&D spending more and more, expect the Compellent product and support to deteriorate. If price is the same, and you can only choose between the two, it's a no brainer to me. EMC all the way. Dell's only competitive advantage is their ability to sell at low or negative margins right now because they're already running a lean business and trying to establish a market foothold.

Compellent operates independently from rest of Dell. Similar to how VMware and EMC storage operate.

All the top players make a good product. Each one can have a great use case. However, people love to pick sides and argue endlessly. Human nature, I suppose.

I always recommend the customer establish their business and technical requirements before going shopping. saves everyone a lot of time and then process is a lot smoother.

If you're struggling to establish your requirements internally, have a consultant come in, do a business and technical audit. Then sit down and use those findings for your response document. Then go looking for vendors.

 

[Conscript]

I have been hearing that from the Dell Haters for over a decade. Still hasn't happened.

well a decade ago they had a successful business model and hadn't acquired Compellent.

Wow... That's some serious FUD.

lol, did I strike a nerve? To get you to spend 25% of your posts on my comment is flattering. But FUD? Hmm, maybe someone with a little more direct insight into their business at the moment?

Compellent operates independently from rest of Dell. Similar to how VMware and EMC storage operate.

They may operate independently, but definitely not even remotely similar to EMC and VMWare. VMWare is an entirely separate company. No matter how independent Compellent is from Dell, they still operate under the same P&L, and ultimately are still affected by the same macro economic conditions of the entire company. Dell had done this before, and I just believe they're going to let this product go to shit. I also think they're a very possible suitor for NetApp, in which case would strengthen my feeling around the compellent product. Personally, I'd probably look at IBM, but wouldn't touch anything from Dell right now with a 10 foot pole.

 

[Brovane]

well a decade ago they had a successful business model and hadn't acquired Compellent.



lol, did I strike a nerve? To get you to spend 25% of your posts on my comment is flattering. But FUD? Hmm, maybe someone with a little more direct insight into their business at the moment?



They may operate independently, but definitely not even remotely similar to EMC and VMWare. VMWare is an entirely separate company. No matter how independent Compellent is from Dell, they still operate under the same P&L, and ultimately are still affected by the same macro economic conditions of the entire company. Dell had done this before, and I just believe they're going to let this product go to shit. I also think they're a very possible suitor for NetApp, in which case would strengthen my feeling around the compellent product. Personally, I'd probably look at IBM, but wouldn't touch anything from Dell right now with a 10 foot pole.

So what is your reasoning that Dell is going to go downhill in the next 5-10 years or sooner?

 

[RobertR1]

They may operate independently, but definitely not even remotely similar to EMC and VMWare. VMWare is an entirely separate company. No matter how independent Compellent is from Dell, they still operate under the same P&L, and ultimately are still affected by the same macro economic conditions of the entire company. Dell had done this before, and I just believe they're going to let this product go to shit. I also think they're a very possible suitor for NetApp, in which case would strengthen my feeling around the compellent product. Personally, I'd probably look at IBM, but wouldn't touch anything from Dell right now with a 10 foot pole.

I'd say Equallogic lineup is likely they most vulnerable line of storage products that Dell holds.

Messing with Compellent and pissing off their enterprise customers is a big no no. If it came down to it, they would spin off or sell Compellent.

I would say "worry about Compellent long term" should be well down on anyone's list of worries when doing a product comparison.

Anyway, looks like Brovane is going with Compellent and the product will serve him well so the thread is more or less over unless we want more Storage warriors from each side to keep it going.

 

[crockjr]

Another happy Compellent customer here. We spent 9 months evaluating EMC, NetApp, and Compellent. I had about 20 meetings with those three vendors and read every "EMC/NetApp/Compellent VS" thread that exists on the web (Google Alerts are great for stuff like that; it's how I saw this thread).

In short we went with Compellent because I like their architecture and I've never researched a product where it was so hard to find a customer who wouldn't buy again. I talked to 5 customers in person (or over the phone) and talked to countless folks online. Compellent is a great product and their support is top notch. Phone Home and their ability to quickly SSH into your array and diagnose problems make support calls go very smoothly. They also have some exciting new features coming out (sorry... NDA; oddly I couldn't find chatter online or I would share a link).

In terms of EMC I ultimately felt that they had a decent solution but it is a bit outdated and overly dependent on cache. They also merely shoe-horned two older product lines into their VNX line. Naturally, they have added some new features, but their snapshot technology is outdated and their user interface needs a major overhaul... all just my opinion of course. To EMC's credit though, they must be doing something right because they have a huge customer base.

One other thing to look for while you reading these threads online. Most of the time EMC employees and vendors come to the rescue of EMC in threads like this (to their credit, they are open about it). Make sure you are finding customers who are willing to talk about EMC, not just folks loyal to EMC because it pays well to work there and/or sell their products (Reddit is a great start).

Feel free to PM me if you would like to chat over the phone; I could speak volumes on this subject!

what kind of performance have you gotten from your compellent san????

 

[cgehring]

what kind of performance have you gotten from your compellent san????

They say a picture is worth a thousand words. In this case it's worth 50,000 IOPS

Overal SAN:

We recently finished our 4th system upgrade, which I mentioned was coming earlier in this thread. It went very well and we are now running on Series 40 64-bit controllers. We ended up adding 16 SSDs (~2.4TB Usable), 48 400GB 15K Drives, and 36 2TB SATA Drives taking the system from 150TB RAW to 225TB RAW.

We have not moved much to the SSD yet, but we have moved a few critical systems and already seeing a HUGE improvement. Latency is almost non-existent. Most noticible was our VDI deployment which is currently only 75 desktops, but we say a 3-4X gain in boot time and login time.

SSD Tier Only:

Performance of spinning disks is a function of physics, and not clever marketing. EMC can't make a 15K drive spin any faster than Compellent can. Make sure you know what performance you need, and buy the correct count of spinning disks to meet that need with room to grow. In other words, either system will perform the same given the same number of spindles. If you can afford it, SSD changes the game a bit, and how each vendor uses SSD is different so depending on your use case one might have a better solution than the other. No 1 SAN is the best for every situation.

At the end of the day for me, the thing that moved me from EMC to CML was really, REALLY good software that makes management and reporting simple and easy.

 

[Jeff7181]

What's the latency look like when you're doing nearly 50k iops?

 

[cgehring]

What's the latency look like when you're doing nearly 50k iops?

Great question. I should have included those numbers as they are very important. Latency does increase with the high IOPS, and we hope we don't stay in the 50K range for very long That's basically our maximum based on our drives and we don't like to hang out at max. Right now we are doing what is called a "RAID Rebalance". I work for a University, and students arrive soon, so we have the rebalance priority cranked up to have it finish before students arrive. When you add disks to a compellent array, you kick off a process that balances the existing blocks evenly across all the new disks. The benefit is every LUN now takes advantage of the newly available IOPS. No pesky LUN Migrations.

We look at latency per tier, and SSD is almost always <1ms.

15K is the busiest tier handling the bulk of the IOPS right now is a very flat line at 7ms read and 9ms write.

7.5K is the variable. 3TB drives do not offer good response times when full. Ours are full, and not yet done rebalancing.... another 55TB to go. We are ranging from 10-25ms read and 30-40ms write. Those will come way down once the rebalance is complete. Until then, we are running them at 100% of what they can do and this is driving up latency.

 

[poopaskoopa]

They say a picture is worth a thousand words. In this case it's worth 50,000 IOPS

Overal SAN:

We recently finished our 4th system upgrade, which I mentioned was coming earlier in this thread. It went very well and we are now running on Series 40 64-bit controllers. We ended up adding 16 SSDs (~2.4TB Usable), 48 400GB 15K Drives, and 36 2TB SATA Drives taking the system from 150TB RAW to 225TB RAW.

We have not moved much to the SSD yet, but we have moved a few critical systems and already seeing a HUGE improvement. Latency is almost non-existent. Most noticible was our VDI deployment which is currently only 75 desktops, but we say a 3-4X gain in boot time and login time.

SSD Tier Only:

Performance of spinning disks is a function of physics, and not clever marketing. EMC can't make a 15K drive spin any faster than Compellent can. Make sure you know what performance you need, and buy the correct count of spinning disks to meet that need with room to grow. In other words, either system will perform the same given the same number of spindles. If you can afford it, SSD changes the game a bit, and how each vendor uses SSD is different so depending on your use case one might have a better solution than the other. No 1 SAN is the best for every situation.

At the end of the day for me, the thing that moved me from EMC to CML was really, REALLY good software that makes management and reporting simple and easy.

When you say overall SAN, is that what's passing through all your controllers, or is that from one controller?

 

[cgehring]

When you say overall SAN, is that what's passing through all your controllers, or is that from one controller?

That is a aggregate of both controllers. Each controller is handling about half that load.

 

[poopaskoopa]

That is a aggregate of both controllers. Each controller is handling about half that load.

Thanks. We run NetApp gears here but I'm always interested in what kind of numbers others getting.

I just looked up series 40. So are you getting those numbers from 2 controllers that have one quad-core processor in each, with 16 SSDs (~2.4TB Usable), 48 400GB 15K Drives and 36 3TB drives split between two of them? At any rate those are pretty high write MB/s numbers.

 

[cgehring]

Thanks. We run NetApp gears here but I'm always interested in what kind of numbers others getting.

I just looked up series 40. So are you getting those numbers from 2 controllers that have one quad-core processor in each, with 16 SSDs (~2.4TB Usable), 48 400GB 15K Drives and 36 3TB drives split between two of them? At any rate those are pretty high write MB/s numbers.

Those numbers only include the upgrade. The system is much bigger than that I wish it were possible to hit those numbers with just those disks. The current config is as follows:

Drive Type Shelves Per Shelf Total Drives
15K-FC 6 16 96
15K-SAS 4 12 48
7.5K-FC/SATA 4 16 64
7.5K-SAS 3 12 36
SLCSSD 1 16 16
Total Drives: 260


Sorry for the awful formating of that table.

 

[Brovane]

I thought I would updated this thread since we went through or EMC CX3-40 to Compellent Migration a couple of weeks ago.

The Compellent we purchased has 60x600GB 15k SAS and 36x2TB 7k SAS disks in it. On the CX3 array we had a 2-Node FS cluster (2k3),3-Node SQL Enterprise cluster (2k3), 2-Node SQL Standard Cluster (2k3), and a 5-Node ESX Cluster and 2-Node ESX Cluster. We paid Dell to migrate up to 15TB of data from the Windows Machines into the Compellent array.

The migration was interesting. Essentially the Compellent array was setup in zoning so the CX3 array could see it as a host. We would take server clusters offline. We would remove the LUN's from the current storage group in the EMC array and assign the LUN to the new storage in the CX3 array that was created for the Compellent. You would then go into the Compellent and the LUN's in the COmpellent storage group on the CX3 array would appear as external disks on the Compellent array. We would then copy those external disks to a internal Compellent volume. We could move 4 LUN's over at a time using this method.

One hitch that we did experience was disk usage/ versus disk size for migration times. In my previous experience with disk migration at the bit level on a SAN, the SAN doesn't know if there is data or not and it is essentially moving bits over and it moves all the bits over. So if you have a 2TB volume to move over with 1TB of usage you still need to move over 2TB of data. Dell told me several times during the project meetings leading up the migration that the Compellent array only copied over bits that had data on them and empty space was simply discarded. So for the 1st test I had a 6TB EMC volume with 250GB of data. We setup everything and it took around 8-10 hours to move the data over. The Dell team was shocked, I wasn't. Basically it didn't matter if the disk had data on it or not the bits where still moved over. We moved through this and we had migration rates of around 1TB/hour during or outage windows.

Overall everything went fairly smooth. I like that I no longer have to mess with PowerPath. The last couple of weeks the Data Progression is starting to level things out and significant amounts of my data have migrated into T3 storage know. This weekend I have using storage motion moved over all most all the ESX guests into the Compellent array. Hopefully in the next couple of weeks this can be all finished up and I can decommission and un-rack the CX3 array. We ordered 2xFS8600's with the array purchase however the FS8600's haven't been officially released yet.

Some caveats.

#1- The data migration wasn't as Dell described since it also moved over empty space on the volumes. During the setup and testing phase the storage architect assigned to or account was onsite making sure everything went smoothly. He is still trying to track down why this didn't work as I was originally told. He is telling me that depending on who he talks to he gets different answers.

#2- The install team/migration from Dell was 1 person. So during the weekend when we had or outage windows we had some really long days. I was using myself and 2 other engineers to make sure that somebody was onsite at all times during the migration. Dell had one person. In about a 48-hour period the Install Engineer got about 4 hours of sleep so by Sunday morning he was fading quickly. Luckily we didn't run into any issue on Sunday or he would have probably been useless. I told or Dell Sales team that this wasn't good practice and isn't very practical to have one install engineer covering all this.

We have 3 more CX3 array's that we are migrating off of next year into Compellent so hopefully we can use what we learned to make the process next year even smoother.

 

[Jeff7181]

I'm not sure I understood what you said, but it sounds like what you did was have the Compellent controllers/heads/array (whatever you want to call it) virtualize the EMC storage. We did the same thing where I work, but with NetApp virtualizing an EMC SAN.

Here's a question for you, I assume Compellent has some sort of mirroring or replication feature. And if I understand correctly, when you connected the CX3 to the Compellent, you were able to access existing LUNs but, but not as a native volume that it could use as a replication source, so you weren't able to simply mirror/replicate the LUNs to native Compellent disk?

If that's the case, then yeah, you'd either be moving stuff on the file level or block level. A block level copy has no way of knowing what's data and what's white space, so I'd fully expect it to copy "empty" space. Makes me think the person who told you it wouldn't copy white space was mistaken.

 

[Brovane]

I'm not sure I understood what you said, but it sounds like what you did was have the Compellent controllers/heads/array (whatever you want to call it) virtualize the EMC storage. We did the same thing where I work, but with NetApp virtualizing an EMC SAN.

Here's a question for you, I assume Compellent has some sort of mirroring or replication feature. And if I understand correctly, when you connected the CX3 to the Compellent, you were able to access existing LUNs but, but not as a native volume that it could use as a replication source, so you weren't able to simply mirror/replicate the LUNs to native Compellent disk?

If that's the case, then yeah, you'd either be moving stuff on the file level or block level. A block level copy has no way of knowing what's data and what's white space, so I'd fully expect it to copy "empty" space. Makes me think the person who told you it wouldn't copy white space was mistaken.

The Compellent SAN appeared as a host to the EMC array so EMC LUN's could be assigned to the Compellent array and be seen as external disks to the Compellent array. We then did you a Copy of the EMC LUN's into the Compellent array to internal disks. Which would make me think that it was at the block level. I would fully expect it to copy empty space. What was interesting was I had Compellent class this past week and the instructor said the exact same thing that Dell was originally telling me was that it didn't copy white space. I pointed out to him my recent experience and he acted very surprised. I am kind of wondering how all these people could be so wrong about how the migration process works? I mean it wasn't like we where the first migration from EMC to Compellent.

 

[Jeff7181]

The Compellent SAN appeared as a host to the EMC array so EMC LUN's could be assigned to the Compellent array and be seen as external disks to the Compellent array. We then did you a Copy of the EMC LUN's into the Compellent array to internal disks. Which would make me think that it was at the block level. I would fully expect it to copy empty space. What was interesting was I had Compellent class this past week and the instructor said the exact same thing that Dell was originally telling me was that it didn't copy white space. I pointed out to him my recent experience and he acted very surprised. I am kind of wondering how all these people could be so wrong about how the migration process works? I mean it wasn't like we where the first migration from EMC to Compellent.

Maybe there are caveats... for example, maybe only a thin provisioned LUN won't copy white space.

 

[Brovane]

Maybe there are caveats... for example, maybe only a thin provisioned LUN won't copy white space.

I know there are always caveats. I would hope either the Install engineer or someone could identify what those caveats are and be able to let a customer know if the caveats will apply to them. IE only with a thin provisioned LUN will you not copy white space. It isn't like I do SAN migrations every month etc. Which is why we pay for the migration service. I am not a storage engineer. I have to wear many different hats.

 

[Jeff7181]

I know there are always caveats. I would hope either the Install engineer or someone could identify what those caveats are and be able to let a customer know if the caveats will apply to them. IE only with a thin provisioned LUN will you not copy white space. It isn't like I do SAN migrations every month etc. Which is why we pay for the migration service. I am not a storage engineer. I have to wear many different hats.

Yeah, I wasn't all that impressed with some of the technicians that various storage vendors sent out to do installations for a POC I was running. We have two dedicated storage administrators where I work and we prefer to do everything ourselves.

 

[Emulex]

I never understood this part, honestly modern storage systems that i've used (3par/lefthand) are so damn simple that within a half hour of reading through the manual and firing up the webclient it was all done, clustered, etc.

You don't have to be a storage engineer to create a lun or cluster or site.

 

[Brovane]

I never understood this part, honestly modern storage systems that i've used (3par/lefthand) are so damn simple that within a half hour of reading through the manual and firing up the webclient it was all done, clustered, etc.

You don't have to be a storage engineer to create a lun or cluster or site.

Usually with storage equipment the initial setup can be fairly more complicated than the actual day/day administration. Especially when you are in a shop with no dedicated storage admins. Everyone is wearing many different hats. So it is easier to just put the initial setup and migration of the new SAN storage in the purchase quote.

 

[Jeff7181]

I never understood this part, honestly modern storage systems that i've used (3par/lefthand) are so damn simple that within a half hour of reading through the manual and firing up the webclient it was all done, clustered, etc.

You don't have to be a storage engineer to create a lun or cluster or site.

There's a bit more to storage administration/engineering than simply getting a storage system to serve data. But you're right, anyone with an IT background should be able to get a SAN to start serving data, especially if it's functioning more as NAS than SAN.

 

[Brovane]

There's a bit more to storage administration/engineering than simply getting a storage system to serve data. But you're right, anyone with an IT background should be able to get a SAN to start serving data, especially if it's functioning more as NAS than SAN.

Yeah but would the SAN be properly setup? I would be concerned with during the setup of the SAN that something that I didn't know, I didn't know would be missed and it would come back to bite me in the butt later. Especially with SAN gear that is expected will be giving you 4-5 9's of uptime.

 

[Jeff7181]

Yeah but would the SAN be properly setup? I would be concerned with during the setup of the SAN that something that I didn't know, I didn't know would be missed and it would come back to bite me in the butt later. Especially with SAN gear that is expected will be giving you 4-5 9's of uptime.

That's exactly my point. The act of serving data is only a small portion of what it takes to administer or engineer a SAN. For some environments, the fact that the data on the device is accessible is enough, for others you need to take into consideration the type of disks you're using, SSD, SATA, SAS, FC SCSI, etc. You need to decide what RAID level you'll use. You need to decide how many spindles will be in a RAID group. You need to decide how many aggregates you'll need and how many spindles you'll need in each. You need to decide how many spare disks you need. If you have two controllers/heads you need to decide whether they're going to be fault tolerant and whether they'll be active/active or active/passive. You need to decide what protocol clients will use to access data on the SAN. You need to consider whether the data needs to be backed up or not, and if so, how?

All of these things can be left at default or recommended settings. The fact that the SAN is up and serving data doesn't mean it's doing it well or that it's as resilient to failures as it could be or that the integrity if the data is maintained.

There's a reason SAN/NAS/Storage Administrator/Engineer positions exist.

 

[OVerLoRDI]

Oh compellent. That system is incredible and unique. I have little experience with SANs, but my company needed someone to pass the compellent top gun architect certification to maintain our partnership level. It was a bit over my head having little exposure to SANs, but it is an awesome system.