Choosing a SAN solution
We have been looking into two SAN solutions to expand our VMware infrastructure. We currently have 1 VMware host with local storage and it is not enough storage or resources for our needs. In addition the host is running on an older machine at this point and needs to be replaced. We also want to have the benefit of redundancy - hence the reason for the SAN.
Part of this solution entails a second, lower level SAN for our DR site. We've been looking at the EQ 6100 (production) and EQ 4200 (DR) and the VNX5300. Both are coming back to us with their prices but it looks as though the VNX solution will be about $30,000 more than the EQ solution.
My first question is regarding the TCO. Dell says their model is better because there is no "rip and replace". EMC says their model is better because if you need to add a bit more down the road you don't need to buy a whole new SAN. I'm sort of leaning towards the Dell approach.
My second question is, from what I've read online the consensus seems to be the VNX is a more powerful machine potentially and all things being equal, many people are opting for the VNX. However in my situation is there any justification to spend that much more for the VNX?
I appreciate any input and suggestions.
Thanks.
Part of this solution entails a second, lower level SAN for our DR site. We've been looking at the EQ 6100 (production) and EQ 4200 (DR) and the VNX5300. Both are coming back to us with their prices but it looks as though the VNX solution will be about $30,000 more than the EQ solution.
My first question is regarding the TCO. Dell says their model is better because there is no "rip and replace". EMC says their model is better because if you need to add a bit more down the road you don't need to buy a whole new SAN. I'm sort of leaning towards the Dell approach.
My second question is, from what I've read online the consensus seems to be the VNX is a more powerful machine potentially and all things being equal, many people are opting for the VNX. However in my situation is there any justification to spend that much more for the VNX?
I appreciate any input and suggestions.
Thanks.
Andrew Hancock (VMware vExpert PRO / EE Fellow/British Beekeeper)
have you looked at any other solutions NetApp?
ASKER
Not in depth just read about it online a bit. Most of our hardware is primarily Dell, and CDW being one of our large vendors is trying to sell us the VNX.
Unless there is a very significant reason that a different product is vastly better for us I'd rather not start from the ground up again at this stage as I really need to make a decision sooner than later.
Unless there is a very significant reason that a different product is vastly better for us I'd rather not start from the ground up again at this stage as I really need to make a decision sooner than later.
If you are familiar I would stick with Dell.
Given the information youve posted, an EMC VNXe3150 may be more appropriate for your needs.
The VNX5300 is certainly a tonne more powerful than the EqualLogics box and offers probably the best VMWare integration among all the storage vendors. It comes closest to giving you a single pane of glass for management and provisioning from vCenter Server with the EMC management plug-ins. The VNX flash drive integration also gives awesome performance at low cost. For the future, if you intend to replicate to a DR site, EMC RecoverPoint is the best replication technology available in the market today. EqualLogic can't compete on that front.
Finally, the forklift upgrade is a red herring. The VNX5300 could expand to something like 250TB before you had to replace the controllers to support more drives. And from what you've posted, that doesn't seem likely. :)
The VNX5300 is certainly a tonne more powerful than the EqualLogics box and offers probably the best VMWare integration among all the storage vendors. It comes closest to giving you a single pane of glass for management and provisioning from vCenter Server with the EMC management plug-ins. The VNX flash drive integration also gives awesome performance at low cost. For the future, if you intend to replicate to a DR site, EMC RecoverPoint is the best replication technology available in the market today. EqualLogic can't compete on that front.
Finally, the forklift upgrade is a red herring. The VNX5300 could expand to something like 250TB before you had to replace the controllers to support more drives. And from what you've posted, that doesn't seem likely. :)
ASKER
From what I've read the VNXe is not a good product - even EMC didn't really want to use it as an option for us as it would be a very tight fit performance wise. The plan is to do replication right away - one of the main purposes of the solution.
My environment will not be quite so big, 2-3 hosts at the primary location and 2 hosts at the DR location. Like I said previously, my understanding is that the VNX product is probably "better" the question is more is it worth the extra cost?
My environment will not be quite so big, 2-3 hosts at the primary location and 2 hosts at the DR location. Like I said previously, my understanding is that the VNX product is probably "better" the question is more is it worth the extra cost?
For RecoverPoint - yes. Absolutely worth every penny. It works a bit like a TiVo for your data, letting you go back to any point In time for recovery. RecoverPoint also does WAN compression and de-duplication so can save a motza on WAN costs. The VNX also has orchestration software (AppSync and Replication Manager) so you can automate application consistent checkpoints with the array and VMware and Windows. .
I'd expect you'd save the extra cost of the EMC kit in bandwidth savings.
ASKER
We currently have 100 MB dedicated fiber at our production site. Our DR site has 100 MB shared, but no where near as busy, so as far as bandwidth is concerned I'm not overly worried. That wouldn't be my overriding reason to choose one over the other.
Lets run the numbers.
How many TB in your production environment?
Do you have a feel for the rate of change?
What is the maximum amount of data you can afford to lose if you have an outage?
How long can you afford your systems to be down?
How many TB in your production environment?
Do you have a feel for the rate of change?
What is the maximum amount of data you can afford to lose if you have an outage?
How long can you afford your systems to be down?
ASKER
Both Dell and EMC are proposing about 10TB based on our current usage.
We need about 1400 IOPS.
We can't really afford to lose any important data at all.
We can't afford our system to be down for any length of time at all. We have people oversees that login to our network even during off hours.
I've read about a difference between Dell and EMC regarding the controllers on the SAN, in regards to being active/active or active/passive.
Do you know anything about this?
We need about 1400 IOPS.
We can't really afford to lose any important data at all.
We can't afford our system to be down for any length of time at all. We have people oversees that login to our network even during off hours.
I've read about a difference between Dell and EMC regarding the controllers on the SAN, in regards to being active/active or active/passive.
Do you know anything about this?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Wow!!
thanks for all the info.
sorry, my numbers may have been a bit misleading. they are calling for 10 TB, but in reality right now we're only using about half. Also from the 5 TB, most of it is a file server with a low rate of change. our exchange and SQL servers are where our high rates of change comes from, but there is less than 1 TB of this kind of really active data.
How does this affect things?
thanks for all the info.
sorry, my numbers may have been a bit misleading. they are calling for 10 TB, but in reality right now we're only using about half. Also from the 5 TB, most of it is a file server with a low rate of change. our exchange and SQL servers are where our high rates of change comes from, but there is less than 1 TB of this kind of really active data.
How does this affect things?
The rate of change derived from the measured 1400 IOPS is the one I'd be working with.
Everything comes back to the business requirements. If you need as close as possible to zero data loss and minimal downtime, Another technology to consider is building a SQL 2012 always-on cluster with two nodes in production and one in DR. Given the scale, You could also consider putting your DR site in Amazon or Rackspace or OpSource cloud. I'd certainly investigate a cloud option for web-facing servers.
Also: Vnx has a built in file server so you can save yourself a Microsoft license and get storage based file system snaps.
Everything comes back to the business requirements. If you need as close as possible to zero data loss and minimal downtime, Another technology to consider is building a SQL 2012 always-on cluster with two nodes in production and one in DR. Given the scale, You could also consider putting your DR site in Amazon or Rackspace or OpSource cloud. I'd certainly investigate a cloud option for web-facing servers.
Also: Vnx has a built in file server so you can save yourself a Microsoft license and get storage based file system snaps.
ASKER
I think I now figured out why the Dell pricing is substantially less than the VNX solution.
In the EQ they are using 24 x 900 GB 10k SAS drives for the PS6100X and the 12 x 2TB 7200 NL SAS Drives for the PS4100E.
While for the production VNX5300 EMC is using 22 x 300GB 15k SAS drives, plus 6 x 2TB 7200 SAS drives and at the DR site VNX5300 they are using 20 x 300GB 10k SAS plus 6 x 2TB 7200 SAS.
Now the VNX requires a few more IOPS for the Recover point software to run on, but it still seems to me that I am not really comparing apples to apples here. And if I were to have Dell match the IOPS of EMC or EMC match Dell's IOPS then pricing would be a lot closer.
On a side note, my reseller is reccomending Cisco servers for hosts to go with the EMC SAN. What are your thoughts on Cisco servers?
In the EQ they are using 24 x 900 GB 10k SAS drives for the PS6100X and the 12 x 2TB 7200 NL SAS Drives for the PS4100E.
While for the production VNX5300 EMC is using 22 x 300GB 15k SAS drives, plus 6 x 2TB 7200 SAS drives and at the DR site VNX5300 they are using 20 x 300GB 10k SAS plus 6 x 2TB 7200 SAS.
Now the VNX requires a few more IOPS for the Recover point software to run on, but it still seems to me that I am not really comparing apples to apples here. And if I were to have Dell match the IOPS of EMC or EMC match Dell's IOPS then pricing would be a lot closer.
On a side note, my reseller is reccomending Cisco servers for hosts to go with the EMC SAN. What are your thoughts on Cisco servers?
ASKER
So I just had a call with Dell and I laid out my discrepancies and they said that in regards to my DR site, they under specked it because they thought from our conversations that I was only planning on using a few servers in a DR situation - they are now trying to salvage things by upping the DR site to a 6100 for "only a few bucks more".
Regarding the production site, they claim that the way the EQ does its storage in a "virtual" manner so that for every vm we create we are getting the IOPS of every disk is a lot better. As opposed to EMC who uses the "Silo" method which is much more inefficient and which is why they need to compensate with more IOPS.
Regarding the production site, they claim that the way the EQ does its storage in a "virtual" manner so that for every vm we create we are getting the IOPS of every disk is a lot better. As opposed to EMC who uses the "Silo" method which is much more inefficient and which is why they need to compensate with more IOPS.
UCS kit is awesome!! UCS is the best blade server infrastructure available on the market today.
I love vendor FUD - some of the rubbish they come up with is awesome.
The original EqualLogic solution for your DR site would have broken if you ever had to fail over to it. The EMC arrays looks to be properly specced for your environment.
The claim about IOPS is silly. I'd expect that the EMC array has been sized using storage pools - which uses every disk. It's also worth noting that the inverse of Dell's claim is that you can't isolate competing workloads because everything is sharing the same spindles and that can be bad. For example, you cant keep database and logs separate on the Dell box.
I love vendor FUD - some of the rubbish they come up with is awesome.
The original EqualLogic solution for your DR site would have broken if you ever had to fail over to it. The EMC arrays looks to be properly specced for your environment.
The claim about IOPS is silly. I'd expect that the EMC array has been sized using storage pools - which uses every disk. It's also worth noting that the inverse of Dell's claim is that you can't isolate competing workloads because everything is sharing the same spindles and that can be bad. For example, you cant keep database and logs separate on the Dell box.
ASKER
It sounds like the VNX series is more comparable with Compellant which is much more expensive than the VNX. If both the Equalogic and the VNX were the same price I think it would be a no brainer - however being the sizeable difference in price, I don't know if my environment requires the VNX solution and spending the extra money which is why I am still back and forth on this.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Yes, I forgot about that, I will definitely look into that some more. When asked Dell about about recoverpoint compared to their replication software, they didn't really have anything to say.....basically no comment.
ASKER
Does the concept that Dell tried to sell me on regarding the EQ having "virtualized" storage as opposed the VNX's "silo" storage, hold any water? I understand its a sales pitch, like the "forklift upgrade", but does it actually mean anything is it really true?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
So like I thought, the only thing that the EQ has going for it is cost and simplicity in setup (which can be an advantage or disadvantage). As far as functionality it sounds like the VNX can pretty much do anything a VNX can plus. I guess it will come down to these things.
It's worth looking at the cost of ownership over three years. Don't forget to factor in WAN bandwidth costs for replication as this is likely to be one of the most significant on-going running costs. If you can save $$$ here, you'll be way ahead of the game.
ASKER
Dell just got back to me this morning that they will give me the PS6100E with 24 x 1TB 7200 NL SAS drives for the same price as they were giving me the 4100 with 12 drive earlier. That's a significant upgrade for the DR site. They are trying really hard - I guess it will come down to price and simplicity vs, power, performance, replication, etc...
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
You might also want to use your DR kit as a test and development environment or to balance your workload across both sites so you have active-active data centers (worth thinking about - half your workload will survive worst case disaster simply because its running Ina different phyiscal location). In all these scenarios you need adequate performance in both data centers.
Hi rivkamak
Any news? :-)
Any news? :-)
ASKER
We're still discussing it internally, and have not yet made a decision. I thank you for all your help and information.
Thanks! Glad I could help.
I'm interested to hear what you decided to go with?
I'm interested to hear what you decided to go with?
ASKER
We ended up going with the Dell Equalogic. Primarily due to the simplicity, our familiarity and the cost.
I'm glad you agreed with my suggestion! as per http:#a39337789