Exchange 2010 beta & high availability strategies

Ewan — Wed, 15 Apr 2009 12:45:00 GMT

Today, the Exchange team released details of Exchange 14, now to be known as Exchange Server 2010. [download here]. There’s plenty of new stuff in the box, but I’m just going to look at one: high availability & data replication.

[My previous missives on Exchange 2007 HA are here, here and here]

There are some interesting differences between 2007 and 2010, particularly in the way databases are handled and what that means for clustering.

THERE IS NO SINGLE COPY CLUSTER ANY MORE.

Single Copy Clusters, or the traditional way of deploying Exchange onto a Windows Cluster with several nodes sharing a copy of the data held in a central SAN, have quite a few downsides … like there being that Single Copy, or the fact that the storage hardware is typically complex and expensive.

There are other pretty major changes, like storage groups going away (it’s just a database now, a move that Exchange 2007 previewed by the advice that you should only have a single DB per SG), or the fact that databases are now the unit of failover (rather than the whole server…), or the ability now to install multiple roles on servers providing high availability – so you could deploy highly available, clustered/replicated environment to a small number of users, without having lots of boxes or VMs.

Oh, Local Continuous Replication goes away too…

Well, reading the documentation explains a bit more about how Exchange 2010 will change the way that high availability can be achieved – no more the need for a MSCS cluster to be set up first should make it simpler, for one. From that site:

Changes to High Availability from Previous Versions of Exchange

Exchange 2010 includes many changes to its core architecture. Two prominent features from Exchange 2007, namely CCR and SCR, have been combined and evolved into a single framework called a database availability group (DAG). The DAG handles both on-site data replication and off-site data replication, and forms a platform that makes operating a highly available Exchange environment easier than ever before. Other new high availability concepts are introduced in Exchange 2010, such as database mobility, and incremental deployment. The concepts of a backup-less and RAID-less organization are also being introduced in Exchange 2010.

In a nutshell, the key aspects to data and service availability for the Mailbox server role and mailbox databases are:

Exchange 2010 uses an enhanced version of the same continuous replication technology introduced in Exchange 2007. See the section below entitled "Changes to Continuous Replication from Exchange Server 2007" for more information.

Storage groups no longer exist in Exchange 2010. Instead, there are simply mailbox databases and mailbox database copies, and public folder databases. The primary management interfaces for Exchange databases has moved within the Exchange Management Console from the Mailbox node under Server Configuration to the Mailbox node under Organization Configuration.

Some Windows Failover Clustering technology is used by Exchange 2010, but it is now completely managed under-the-hood by Exchange. Administrators do not need to install, build or configure any aspects of failover clustering when deploying highly available Mailbox servers.

Each Mailbox server can host as many as 100 databases. In this Beta release of Exchange 2010, each Mailbox server can host a maximum of 50 databases. The total number of databases equals the combined number of active and passive databases on a server.

Each mailbox database can have as many as 16 copies.

In addition to the transport dumpster feature, a new Hub Transport server feature named shadow redundancy has been added. Shadow redundancy provides redundancy for messages for the entire time they are in transit. The solution involves a technique similar to the transport dumpster. With shadow redundancy, the deletion of a message from the transport database is delayed until the transport server verifies that all of the next hops for that message have completed delivery. If any of the next hops fail before reporting back successful delivery, the message is resubmitted for delivery to that next hop. For more information about shadow redundancy, see Understanding Shadow Redundancy.

SMSE – a System Center light hidden under a bushel

Ewan — Thu, 26 Mar 2009 16:47:42 GMT

SMSE – pronounced (in the UK at least) as ‘Smuzzy’, short for Server Management Suite Enterprise – is a licensing package from Microsoft, which can be an amazingly effective way to buy systems management software for your Windows server estate.

If you’re planning to virtualise your Windows server world, then SMSE is something of a no-brainer, since buying a single SMSE license for the host machine allows you to use System Center to manage not just the host but any number of guest (or child) VMs running on it.

Combine that with the license for Windows Server 2008 Datacenter Edition, which allows unlimited licensing for Windows Server running as guests, and you’ve got a platform for running & managing as many Windows-based applications servers as you can squeeze onto the box, running on any virtualisation platform.

System Center is the umbrella name given to systems management technologies, broadly encompassing:

Configuration Manager (as-was SMS, though totally re-engineered), which can be used for software distribution and “desired configuration state” management … so in a server example, you might want to know if someone has “tweaked” the configuration of a server, and either be alerted to the fact or maybe even reverse the change.
Operations Manager (or MOM, as it was known before this version), performs systems monitoring and reporting, so can monitor the health and performance of a whole array of systems, combined with “management packs” (or "knowledge modules” as some would think of them) which tell Ops Mgr how a given application should behave. Ops Mgr can tell an administrator of an impending problem with their server application, before it becomes a problem.
Data Protection Manager – a new application, now in 2nd release, which can be used either on its own or in conjunction with some other enterprise backup solution, to perform point in time snap shots of running applications and keep the data available. DPM lets the administrator deliver a nearer RTO and more up to date RPO, at very low cost.
Virtual Machine Manager – a new server, also in 2nd release, which manages the nuts & bolts of a virtual infrastructure, either based on Microsoft’s Hyper-V or VMWare’s ESX with Virtual Center. If you have a mixture of Hyper-V and VMWare, using VMM lets you manage the whole thing from a single console.

It’s easy to overlook managing of guests in a virtualised environment – the effort in doing such a project typically goes into moving the physical machines into the virtual world, but it’s equally important to make sure that you’re managing the operations of what happens inside the guest VMs, as much as you’re managing the mechanics of the virtual environment.

I’ve used a line which I think sums up the proposition nicely, and I’ve seen others quote the same logic:

If you have a mess of physical servers and you virtualise them, all you’re left with is a virtual mess.

Applying the idea of SMSE to a virtual environment, for one cost (at US estimated retail price, $1500), you get management licenses for Ops Manager, Config Manager, VMM and DPM, for the host machine and all of its guests.

Think of a virtualised Exchange environment, for example – that $1500 would cover Ops Manager telling you that Exchange was working well, Config Manager keeping the servers up to date and patched properly (even offline VMs), VMM managing the operation of the virtual infrastructure, and DPM keeping backups of the data within the Exchange servers (and maybe even the running VMs).

Isn’t that a bargain?

See the FAQ for SMSE for more information.

The Electric Wand : Infrastructure

Exchange 2010 beta & high availability strategies

SMSE – a System Center light hidden under a bushel