Three node Single Copy Cluster A/A/P

Three node Single Copy Cluster A/A/P (Full Version)

All Forums >> [Microsoft Exchange 2007] >> High Availability

Message

sbruton -> Three node Single Copy Cluster A/A/P (4.Oct.2007 9:01:23 AM)
Ok, As were aware Active/Active is no longer supported in Exchange 2007, Therefore its Active/Passive or Active/Active/Passive Here in lies my question, If I build and I was going to in order to find out, but don’t have enough memory for the virtual server solution, But if I build an Active/Active/Passive Exchange 2007 SCC, using three servers Whereby I want to place 5000 users in total, with say 2500 on each of the active nodes and have the passive as standby, Node1 Active EXCMBX01 SG1 & SG2 Node2 Active EXCMBX01 SG3 & SG4 Node3 Passive EXCMBX01 Do I end up with One Exchange Server name such as EXCMBX01 with the two active nodes hosting part of the storage group each Or do I have to configure two separate EXCMBX01 and EXCMBX02 with 2500 users on each server with the passive as the failover for either of the two servers ? Node1 Active EXCMBX01 SG1 & SG2 Node2 Active EXCMBX02 SG1 & SG2 Node3 Passive EXCMBX01/02 What do we think, has anyone built a three node A/A/P solution ? Thanks Stephen

John Weber -> RE: Three node Single Copy Cluster A/A/P (4.Oct.2007 11:45:41 AM)
First off, CCR: I do not think you can do three nodes. A/P is all you get. Second, if you are going to be using CCR, and you have 5000 users, I would use more databases; which, in CCR, means an SG per DB. -jmw

neilho -> RE: Three node Single Copy Cluster A/A/P (4.Oct.2007 1:42:00 PM)
True John, but he's asking about SCC and not CCR. [;)] Stephen - you'd end up with two clustered mailbox servers that your clients can connect to. The passive node will be available to take over if one of the clustered mailbox servers fails. Microsoft's current recommendation is to have as many passive nodes as you have as active nodes, i.e. you'd have an A/A/P/P solution. That's not to say that A/A/P won't technically work because it will. However, in the event of a failure, you then have no spare nodes until you get the problem fixed.

sbruton -> RE: Three node Single Copy Cluster A/A/P (8.Oct.2007 7:56:37 AM)
Thanks all for the replies, yes it is a Single Copy Cluster SCC, and I was aware of the A/A/P/P preference, but getting someone to cough-up for two standby servers is not that easy. That said I have 5000 users so 2500 on each. I just need to finalize my storage design now, I have opted for 5 SG on each server with 1 DB in each. Small 100Mb mailboxs but heavy daily usage 50 in 125 out. Im allowing for 5% for indexing and 110% for isenteg. Any comments Regard Stephen

neilho -> RE: Three node Single Copy Cluster A/A/P (9.Oct.2007 8:59:35 AM)
Other sizing factors: 1. 20% free space on the LUNs at all times, just so the LUNs aren't operating near capacity. (MS recommendation) 2. Deleted item retention factor in the storage allocation 3. Deleted mailbox retention factor 4. What backup failure tolerance are you specifying? i.e. additional transaction log space in case the backups fail. 5. Will any mailboxes be moved each week to rebalance the stores? If so, transaction logs will be created.

sbruton -> RE: Three node Single Copy Cluster A/A/P (9.Oct.2007 9:13:00 AM)
Thanks, I think I have taken most of that into account, see inline Other sizing factors: Mailboxs are sized for 100Mb each, small I know but standard policy, however it is heavily utilized with expectations of 50 mails in and 100 out per user per day. 1. 20% free space on the LUNs at all times, just so the LUNs aren't operating near capacity. (MS recommendation) I have allowed additional space for this and an additional 5% for indexing 2. Deleted item retention factor in the storage allocation, We have decided on an 8 day deleted item retention policy, 5 storage groups, full backups Monday - Friday, one storage group per night, one database per storage group, so transaction logs flushed every 7 days. 3. Deleted mailbox retention factor, deleted mailbox will be left at 30 days, only 100Mb so not too bad 4. What backup failure tolerance are you specifying? i.e. additional transaction log space in case the backups fail. Were covered for this, the standard NTbackup is been used to back up to disk, with NetBackup then backing up this files 5. Will any mailboxes be moved each week to rebalance the stores? If so, transaction logs will be created. I was aware of the hit mailbox moves will cause, there will be very little movement but that said at 100mb it will take a lot of mailbox moves to generate suffucient logs to fill the tranaction logs. I am concerned slightly over the LUNS however? I will have the following drive specs C & D: are 2146gb mirrored, e: is also 2146 Mirrored C:\os D:\Program Files E: Spare SAN H: SG1 Raid 5 N: LG1 Raid 1 I: SG2 Raid 5 O: LG2 Raid 1 J: SG3 Raid 5 P: LG3 Raid 1 K: SG4 Raid 5 Q: LG4 Raid 1 L: SG5 Raid 5 R: LG5 Raid 1 As this drive letters will be the same on both clustered servers will this cause a problem with the standby server? Thanks Stephen

brota -> RE: Three node Single Copy Cluster A/A/P (2.Nov.2007 3:04:33 PM)
Does anyone have any info on how to install/config the cluster like this? Like how you would configure the SAN and do you choose active node on 2 of the servers during the install? Any info would be very helpful Thanks Brian

Page: [1]