Software

Three reasons why our college is upgrading to Exchange Server 2010

Scott Lowe talks about three reasons why upgrading to Microsoft Exchange Server 2010 makes sense for Westminster College.

Westminster College will be upgrading from Microsoft Exchange Server 2007 to Microsoft Exchange Server 2010 over the holidays. To many IT pros, this will seem like an early, aggressive upgrade, but the product simply has too many compelling features for us to wait.

Our upgrade window is either the holiday break or next summer, but we're counting on three key Exchange 2010 features to start helping us address business and technical issues that we face. Exchange 2010 brings much more to the table than these three simple items, but these items stood out as current drivers for us.

#1: Reduced I/O

Although I don't have much in the way of stats regarding our existing Exchange Server 2007 system's I/O footprint; I do know that our 1,400 or so Exchange mailboxes are stored on an EMC AX4 iSCSI storage array. On that same array, we house all of our SQL databases and a number of virtual machines, which include our new print server, file server, and other lines of business applications. Although the AX4 has performed admirably, I can imagine eventually hitting an overall I/O limit.

One major benefit of Exchange 2010 is that it further reduces the product's I/O footprint by up to 50%, according to Microsoft; I've even seen claims of 70% reductions.  Further, Microsoft has reengineered Exchange to avoid bursting disk writes, which used to keep Exchange from being supported on SATA disks.  With the reengineered disk write methodology, Exchange 2010 can support SATA disks, which leads to a much lower TCO on the product if we go that route.

Edit: We are gathering the necessary performance stats in order to better understand how Exchange is currently impacting our storage so that we can make sound storage decisions.

#2: Outlook Web Access improvements

When Westminster made the move to Exchange 2007, one of the primary drivers was the new and improved Outlook Web Access (OWA), which is the access method of choice for our students and by many staff when they are out of the office.  With Exchange 2010, Microsoft has finally decoupled OWA and Internet Explorer by leveling the browser playing field;  OWA 2010 provides the premium OWA experience for users of IE 7, Firefox 3+, and Safari 3+. With Westminster students and staff using all of these browsers and a combination of Windows and Mac machines, cross-browser support for the premium OWA experience is one of the most compelling upgrade enhancements for us.

OWA 2010 adds other new features, including conversation view, which groups messages from a single conversation, making it easier to keep track of the chain. Other clients have supported this capability for quite some time, so it's a welcome addition to Exchange. With many people allowing access to one another's calendars at Westminster, we're also looking forward to OWA 2010's capability for users to view shared calendars and contacts.

#3: Taming email overload with moderated distribution lists

For quite some time, users have complained about the sheer number of "announcement" messages that hit campus mailboxes on a daily basis. Sometimes, two dozen messages will get sent to students, each advertising an event, meeting, or some other function. In addition, the college regularly sends out critical academic information via email, such as course registration information. Unfortunately, sometimes these important messages are ignored in the onslaught of email.

Up to this point, we've been relatively liberal in allowing messages to go to our primary lists, but the volume has been growing, and we risk devaluing what is a powerful communications channel. The ease with which messages get to peoples' inboxes is a double-edged sword, too. For instance, messages are frequently sent with incorrect or missing key information, such as dates and times. Also, a second (and sometimes third) message is sent clarifying what should have been correct the first time.

Exchange 2010 adds the ability to create moderated distribution lists, which means that users can still email the lists, but the messages have to be approved by a third party before the messages are released to mailboxes. While not our intended solution by itself, Exchange 2010's moderated distribution list capability is a key component in our efforts to reign in messaging and keep email communication relevant on campus.

Cost is an important factor

Because we're on a Microsoft Campus Agreement, Exchange 2010 is provided as a part of our agreement. Our Exchange 2010 implementation, with the exception of the Unified Messaging component, will be installed on existing virtual machines and with our existing SAN, so there is no need to buy additional equipment. Since we'll migrate user mailboxes from Exchange 2007 to Exchange 2010, we won't have additional space needs, either. Once we're fully migrated from Exchange 2007 to Exchange 2010, we'll repurpose the existing Exchange 2007 system as our Exchange 2010 Unified Messaging server.

We plan to begin migrating people to Exchange 2010's Unified Messaging. We started using Unified Messaging in Exchange 2007, but when we upgraded our phone system last August, we implemented the PBX vendor's voice mail system. Exchange 2010 adds some key functionality to Unified Messaging, including a text-based preview of voicemail and, finally, integrated message waiting light indicator capability.  (With Exchange 2007, a phone's MWI light had to be controlled using a third-party product.)

Why not outsource email to Google or Microsoft?

We are asked this question a lot. There are two reasons why we don't outsource our email to Google or Microsoft. Our email infrastructure costs really aren't that bad, so the cost/benefit isn't as great as it is for many other schools. Also, we do a lot of integration with other services and plan to do more, so for us, it makes more sense to keep Exchange in-house.

Related TechRepublic resources

Want to keep up with Scott Lowe's posts on TechRepublic?

About

Since 1994, Scott Lowe has been providing technology solutions to a variety of organizations. After spending 10 years in multiple CIO roles, Scott is now an independent consultant, blogger, author, owner of The 1610 Group, and a Senior IT Executive w...

13 comments
BOUND4DOOM
BOUND4DOOM

I would like to hear more on why not to outsource. We have been debating this decision internally as well. For a global company it make more sense to go to Google. Instant translation, ease of use. Open to all mobile devices. Built in email archiving and so on. All of this makes this a very attractive option. So I would like to hear more about your argument why you made this decision VS others. For us for example server cost money housing them in the data center cost money, our own spam and virus filtering cost money, blackberry servers cost money. So outsourcing for one price with a bunch of additional features with less servers and maintenance is getting more and more appealing. So I would be interested to here more in the process you used to make this decision.

ilya.shick
ilya.shick

Microsoft vs Google. Finally I found a proper answer in this article : Exchange is a part of integration services. Not to many tech folks understand a complex of existing and future services Plus Google Email system failed more than four times for a few month Contrad

AstroCreep
AstroCreep

What are you going to do in regards to the removal of Single Instance Storage, Scott? That is going to be a large hurdle for my company to overcome. Once we have an e-mail storage policy in place it may not be so bad, but most at the executive level are digital pack-rats, with mailboxes over 3GB in size. What are you planning to do about it?

stephen.sandifer
stephen.sandifer

Not because there's a justifiable reason for it. I'm not opposed to IT decisions made for this reason, but I'd sure like to listen in when you pitch this to your financial controller. #1: Ridiculous. You have no real stats from your EMC and you're using marketing numbers from Microsoft. You're using an iSCSI array so the SATA issue isn't relevant. How in the world are you doing capacity planning if you don't know what your EXISTING Exchange box is doing to the storage array? #2: Silly. So you get Firefox and Safari OWA support. You're really going to go through a major upgrade to please the fraction of users who are OWA-centric, and the fraction of THEM who don't use IE? #3: Fix. This. Now. There is zero reason for anyone to send to an all-hands distribution list without a trusted person's say-so. This behaviour is the biggest black-eye IT can get--annoying its customers! Set the distribution groups to only receive from a set address, task that role-holder the 15 minutes a day it'll take to moderate submissions, and be done with it. You're the CIO! Set the policy! #4(cost) I'll assume you've already tested Exch2010's Unified Communications and you have a blood-signed promise from your integrator/PBX vendor that it'll scale. You do, don't you? And you've planned for the additional load on the SAN . . . nope, no stats. And an MWI light? Really? So the takeaway for this point is an MWI light on the phone and voice-to-text processing for voicemail? 2010 sounds like a great product. But this article is a lousy justification for a real-world IT project.

james_smale
james_smale

Interested to know if Google Apps was evaluated as part of this decision. They have a compelling story relative to cost savings, collaboration and innovation.

The 'G-Man.'
The 'G-Man.'

the Microsoft Campus Agreement. Tell me, without that, would you still upgrade?

james.peluso
james.peluso

Hey Scott, Nice article I also follow your blog. I work in Education as well and it's nice to know my group isn't the only one considering going to Exchange 2010 right away. We are mainly driven by the new features of OWA. Jim

tom.marsh
tom.marsh

The main drawbacks are: 1) Liability: If Google gets compromised and your customer's data gets compromised your customer might sue Google... but they're also pretty likely to sue you since you are the one who gave their data to Google. 2) Integration/Authentication: Either you spend an arm and a leg on a SAML based integration product, or you suffer an unpleasant fate: Batch integration. Or, worse, NO integration at all--and your users can juggle passwords indefinitely. Unfortunately, Google does not (to date) support the use of ADFS for integration--so even if you're on the Microsoft Academic license (where Server Enterprise licenses are mucho-cheapo) you can't use the Identity Federation product you've already paid for. 3) Disaster Recovery: Google recently announced that they had "lost" (unrecoverably) email for several thousand user accounts. What would happen if they "lost" your company's data? This is possible with your own systems, of course, but when the systems belong to YOU, you can engage in a little responsible oversight. When the servers belong to Google, you are only able to take their word for it. 4) Regulatory Compliance: Google is pretty great, but don't count on any regulatory agencies your company needs to work with buying in without a fight. This will cost a considerable amount of money (in man-hours) and is part of the equation. 5) Accountability: Unless your company is rich enough make a credible legal threat to Google, there simply isn't any accountability whatsoever. Good luck getting damages out of them--your "user agreement" doesn't provide for anything beyond a refund if they fail to deliver. What this means is, essentially, if Google screws up, YOU will bear the brunt of it. 6) Their support is a joke. Non-existent would be a better label. I migrated the students of a college to Google Apps for Education. Google supports literally NOTHING, unless you can prove that their (mostly undocumented) API has a bug in it. Support is not included and, at least when I was involved with it, simply didn't exist. The make it nearly impossible to contact support, and you can't enter a ticket without giving a secret code that is only available through your web-interface. In short, it isn't as "perfect" as it might seem on the surface. Certainly, there are plenty of positives, but you can't just cross the cost of Email and support off your budget and move-on. Some of what you save will be burned in integration costs or support costs. Given the potential legal liabilities, hidden costs, and total lack of support from Google, my own recommendation would be to not even attempt it unless your company is SOO HUGE that you can take on Google, or so small and broke that nobody would waste their time suing you.

Scott Lowe
Scott Lowe

A few things to bear in mind: * Exchange 2007 doesn't do SIS on message bodies, either. * Although SIS is going away in Exchange 2010, attachments will be compressed. Obviously, from a space perspective, non-SIS will still require additional space, though. * Fewer mailboxes per DB reduces SIS efficiency anyway. Again, not greatly, but noticeably. Most of our mailboxes are sub-100MB with only a few in the multi-GB range. For those users, we are going to help them reduce their mailbox sizes. We've also recently deployed and are beginning to use SharePoint as a collaboration tool. It's already eliminated some of what would have been emailed attachments. As we move forward, we'll be implementing new policies that make it more difficult to blast large files to the entire campus. We will lower caps on max message size to help control mailbox growth.

Scott Lowe
Scott Lowe

It's mainly the new OWA that I'm after for a variety of reasons - mainly, Firefox support. We have a sizeable Macintosh contingent that we want to bring to parity from a service perspective. For us, that's important; for other, it may not make sense. On number 1, we don't have current figures, but are gathering them as a part of our planning phase in order to make sure that we properly account for storage needs under Exchange 2010. #3 - Easier said than done. I'm not going to make a bunch of policy changes without being able to give people some kind of reasonable alternative. That's what we're working on now - 1) Getting people up to speed on the Sharepoint-based campus calendar we just finished developing; 2) Once they're on board, require them to submit event information through the calendar; 3) Configure the calendar to generate a message to campus each day with that day's activities. There's more to it than I said above. It's not all technical. As for UM, yes, it works with our PBX. We're have a limited initial deployment on this front. The SAN is up for replacement relatively soon, so, yes, we're taking a small risk that we'll have to wait until we can get a more robust SAN for this service. I discovered than an MWI light is more important than I thought... users really want it. If it's reasonable to provide it, great. Scott

knoxbury
knoxbury

I would go a step further and say that this guy is somehow connected with Microsoft. This sounds like a piece straight out of the MS play book disguised as legitimate press.

tom.marsh
tom.marsh

...you have the mountain of cash required to integrate your existing authentication scheme to Google, then its very compelling. Otherwise, its a nightmare of non-matching passwords, batch integration, and endless support headaches over passwords. Also the joy of accounts that "disappear." Oops! It isn't as "cheap" as the "free" label that Google Apps for Education has presently got slapped on it--still worth investigating, but everybody should go into it with open-eyes.

Scott Lowe
Scott Lowe

Obviously, the economics are pretty important and if you can't justify the cost, you don't do the project. Since the CA basically makes this upgrade "free", we'll do it and not think too much about it. If there was a major cash transaction involved, I doubt we'd be making the jump right now since budgets are tight. Scott

Editor's Picks