multi site san solution - hp p4000, Dell Equallogic PS4100X, NetApp FAS2240
Hi all,
we are talking with multiple vendors about a vmware upgrade to our two sites. The sites are 300-400 metres apart and I hope to link them together with some fibre before the vmware hardware/software upgrade.
The vendors have proposed solutions that are HP Lefthand, Dell Equallogic and Netapp. My experience is fairly limited in the San space so I thought I would get your opinions on this.
I have seen an HP solution and it looked good to me. From what I understand if we went with the HP solution we would have a solution that is true High Availability where several hardware pieces could fail without having a dramatic impact on the whole. In addition we would be using all of the pieces of hardware (such as the sans) rather than waiting for a san to fail before manually bringing up the secondary one. Because we are using essentially 2 sans instead of 1 active san with another solution (such as Netapp) this to me sounds ideal.
So here are my doubts:
The NetApp solution will cost more (I'm guessing +40% at this point for an active passive solution - but it would be fibre and not iscsi), and the vendor pushing their product specialises in vmware projects. They claim they could offer an HP solution and make a higher profit off of it, but they believe it's the wrong solution.
The Equalogic solution I just don't know enough about at this point (I'm trying to sift through reviews, but I'm finding it difficult to sort through opinions and facts). The cost looks like it will be more than the HP solution as it's roughly the same with slower drives at this point. I'm yet to determine if the Dell solution is an active active cluster or if it's active passive, but I suspect the later is the case.
It would be really helpful if someone could point to why the HP solution is better or worse in this case money aside.
In addition, I've been thinking of dedicating 4 switches with 10Gb modules for a separate redundant physical iscsi network (assuming we go with HP or dell). Is this just plan overkill, and instead I should be using those switches for the servers and iscsi traffic with a separate vlan for iscsi?
Thanks,
Mike
we are talking with multiple vendors about a vmware upgrade to our two sites. The sites are 300-400 metres apart and I hope to link them together with some fibre before the vmware hardware/software upgrade.
The vendors have proposed solutions that are HP Lefthand, Dell Equallogic and Netapp. My experience is fairly limited in the San space so I thought I would get your opinions on this.
I have seen an HP solution and it looked good to me. From what I understand if we went with the HP solution we would have a solution that is true High Availability where several hardware pieces could fail without having a dramatic impact on the whole. In addition we would be using all of the pieces of hardware (such as the sans) rather than waiting for a san to fail before manually bringing up the secondary one. Because we are using essentially 2 sans instead of 1 active san with another solution (such as Netapp) this to me sounds ideal.
So here are my doubts:
The NetApp solution will cost more (I'm guessing +40% at this point for an active passive solution - but it would be fibre and not iscsi), and the vendor pushing their product specialises in vmware projects. They claim they could offer an HP solution and make a higher profit off of it, but they believe it's the wrong solution.
The Equalogic solution I just don't know enough about at this point (I'm trying to sift through reviews, but I'm finding it difficult to sort through opinions and facts). The cost looks like it will be more than the HP solution as it's roughly the same with slower drives at this point. I'm yet to determine if the Dell solution is an active active cluster or if it's active passive, but I suspect the later is the case.
It would be really helpful if someone could point to why the HP solution is better or worse in this case money aside.
In addition, I've been thinking of dedicating 4 switches with 10Gb modules for a separate redundant physical iscsi network (assuming we go with HP or dell). Is this just plan overkill, and instead I should be using those switches for the servers and iscsi traffic with a separate vlan for iscsi?
Thanks,
Mike
ASKER
Thanks,
Do you happen to have any good data about IOPS between the systems? I found a claim that basically suggested that Equalogic was twice as fast as Hp lefthand. We will have twice as many nodes so that might not be a huge issue, but if it is true I'm wondering why.
The HP san's have 15K sas drives in them and the Equalogic had 10Ksas. I would think that IOPS would be with the Faster drives.
Thanks for the Tip related to the 10Gb switches. We may still go down that route, but I gathered I might be a little OTT.
Do you happen to have any good data about IOPS between the systems? I found a claim that basically suggested that Equalogic was twice as fast as Hp lefthand. We will have twice as many nodes so that might not be a huge issue, but if it is true I'm wondering why.
The HP san's have 15K sas drives in them and the Equalogic had 10Ksas. I would think that IOPS would be with the Faster drives.
Thanks for the Tip related to the 10Gb switches. We may still go down that route, but I gathered I might be a little OTT.
How many IOPS do you require for your virtual servers?
There's an interesting whitepaper at http://content.dell.com/us/en/enterprise/d/dcm/ptech-virtualized-exchange-2010-equallogic-vs-lefthand-fibre-channel that "proves" equallogic to be faster than LeftHand. The title's wrong though, it should have the heading "how we totally misconfigured the LeftHand storage so the Dell beat it". (they used Network RAID 10+1 instead of installing a failover manager by the looks of it).
ASKER
hanccocka,
Honestly, I'm pretty new at looking at requirements like IOPS. I do know we will be running an exchange server for 300 or so staff, perhaps another over 1000 students, plus a pretty intensive proxy server, and eventually some other databases. Along with your typical services DHCP, DNS, AD etc...
What exactly that means in IOPS I don't really know. I do know that our current solution md3000i/md1000i with a mix of data and sas (could be older scsi actually) drives with 3 PE2950 (32GB of ram) just isn't cutting it. We need something better. I would like to shoot for something that would be twice the performance of the md3000i. I gather the any of the above solutions will be a huge improvement. Would you agree?
Honestly, I'm pretty new at looking at requirements like IOPS. I do know we will be running an exchange server for 300 or so staff, perhaps another over 1000 students, plus a pretty intensive proxy server, and eventually some other databases. Along with your typical services DHCP, DNS, AD etc...
What exactly that means in IOPS I don't really know. I do know that our current solution md3000i/md1000i with a mix of data and sas (could be older scsi actually) drives with 3 PE2950 (32GB of ram) just isn't cutting it. We need something better. I would like to shoot for something that would be twice the performance of the md3000i. I gather the any of the above solutions will be a huge improvement. Would you agree?
ASKER
andyalder,
Thanks for the note. I generally take the manufacturers comparisons with a chunk of salt.
So you are a fan of HP Lefthand?
- mike
Thanks for the note. I generally take the manufacturers comparisons with a chunk of salt.
So you are a fan of HP Lefthand?
- mike
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
NetApp and Equallogic heads, will replicate LUNs to other SANs.
How many virtual machines, 10Gb switches may be a little OTT, dedicated 1Gb switches for storage traffic will be fine, but we've not seen much difference in pricing between the two technologies today.