Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations strongm on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Exchange 2010 RTM! 1

Status
Not open for further replies.

58sniper

MIS
Apr 25, 2004
9,152
US
With a vote amongst the internal groups within the Exchange product group, as well as those in the TAP program, Microsoft Exchange 2010 was released to manufacturing today.

As someone who's been running on various Exchange 2010 builds in production for over a year, I can say that this version is by far the most feature rich, high performing, and stable version that the product group has ever released. Organizations looking to upgrade their existing Exchange environments should give 2010 a thorough look, as should orgs looking to migrate from non Microsoft platforms such as Notes, SendMail, and GroupWise.

Kudos to the product group for an outstanding product!

Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
 
Well, yes it did.

I have to say, having run Exchange 2010 for over 6 months now (various buids), that it is increadibly performant. I'm running the RC on SATA now, and it's impressive. If you are considering an upgrade to Exchange 2007, take a look at 2010 as well. It's worth the look.



 
Here are the obvious questions that people reading your responses will ask:

Why is it better than 2007
What does it do that will talk the bean counters into approving 2010 which is untried over 2007 which is now at SP2 and will therefore work. Remember that all MS products aren't stable until SP2...
 
Substantialy lower IOPS, which XMSRE can speak about. Multiple copies of the same database means no need for backups. Seriously? Yes!

You can have all of the typical roles on a server, like Hub Transport, CAS, and Mailbox, and THEN put HA options on - after installation. Neither of those was possible in 2007.

ECP and enhanced OWA give the user more control and better experience in non-MS browsers.

Cross forest functionality is much better.

Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
 
Ok, let's talk about the space this product is good for. For clients that are just about to pull the trigger on new Exchange 2007 deployments, how does the prospect of 2010 change things? For example:

What's the minimum number of servers you can have?

And how many to realize the benefits of multiple database copies (DAG)?

Dave Shackelford
ThirdTier.net
 
In my environment I'm waiting for Exchange 2010 to go out,and to build a redundancy solution.
I wanted to know whether I can do it using only 2 servers, or we'll need again more like in Exchange 2007.
What about archiving option?
 
You can do two servers and have a DAG. That's the fewest number of server & still have HA.

Dave - 1 server. Just like 2007, you can put all of the "typical" roles on a single server.

Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
 
Alfa2008, if you use only 2 servers (each with all the roles - MBX, HUB transport, CAS) then you'll need a hardware load-balancer (NLB doesn't function with this type of DAG setup).

By the way, MS recommend a minimum of 3 servers for your DAG for better resiliency. Then you can cope with both a planned node unavailability (say for patching) and a disk or server failure simultaneously.
 
zbnet, Thank you for clarifications.

As a person without experience in clustering, and NLB, I would like to know what I need all to run full Exchange 2010 redundant.
I assume for 2 servers, I need Windows 2008 Enterprise, and Exchange 2010 Enterprise.
What do I need for the 3rd one playing NLB, Windows 2008 Std,or Ent, Exchange 2010 Std, or Ent.
I hope there will be some books out soon to explain it in more details, too.
 
Let me clarify my previous statement: for servers in a DAG (no matter how many servers you have in the DAG), if they are running CAS as well as MBX then you need a hardware load-balancer to share the client connections to the CAS roles.

To use NLB to share client connections to a pool of CAS servers, those CAS roles must run on Exchange servers that aren't part of a DAG (ie separate servers form the DAGed MBX servers).

All servers running in a DAG must have Windows Enterprise edition.
 
So, if I counted well, in simpliest case, I would need:
2x Windows 2008 Enterprise
2x Exchange 2010 Enterprise
1x Windows 2008 Standard
1x Exchange 2010 Standard

Apparently, it means there is no much difference between Exchange 2007, and Exchange 2010 regarding hardware requirements, meaning number of servers needed.
 
I'm confused about what you're trying to achieve here, and I'm also confused about why you're mixing versions.

The simplest configuration is a single server, running MBX, HT and CAS roles. It can do that with SATA disks, although with a single server you'd probably want to RAID them.

2 servers, both running Win Ent, both running all three roles (MBX, HT, CAS) can be DAGed, but you'd need a hardware load-balancer to share the client load between them.

More nodes (up to 16) can be added to the DAG, as appropriate. As the number of replicated copies of the DBs goes up (as the number of nodes in the DAG goes up), the reliance on disk redundancy becomes less important, so maybe you can run your DAGed servers as JBOD or RAID 0 on SATA and save even more money.

You could avoid the hardware load-balancer and use NLB if you really wanted to (although hardware load-balancing is the proper corporate solution) by having 2 (or more) DAGed MBX servers, and 2 (or more) CAS/HT servers that run NLB to share the client load. But that strikes me as less optimal and more expensive. Hardware load balancers aren't that expensive, after all.
 
Sorry, but I misunderstood you in your first reply. I missed hardware load-balancer.
If I want connect all clients in one server, but to use second one only as a redundant, for clients to connect in case of failure, or regular maaintenace, can it be achieved only with 2 server. I don't have problem with budget, but just want to know, and after that to make right calculations. I don't think we don't need here more than 2 DAGs.
 
You're looking at it wrong. If you have two servers, why not use them all of the time instead of just during a failure? Load balancing makes sense. If you need to take one down for maintenance, or if one fails, you're still using the other. And - each only has to deal with 1/2 the load when both are online.

Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
 
Since I didn't have to do with hardware load balancer, can anybody propose one, which will serve this purpose, and do clustering between 2 of them, if possible, in order to have full redundancy.
 
you don't do "clustering" with a hardware device. Exchange has built in HA via DAG, etc. The hardware load balancer would just load balance connections.

Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
 
I was thinking for a redundant solution for hardware load balancing itself.
 
Yes, you can do that - in fact a fully-redundant solution would indeed include fail-over load balancers.
 
I would be grateful if you could suggest one that could serve the purpose.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top