|
 |
|
General Information
|
Add To My Personal Library |
April 10, 2009
• Vol.31 Issue 12
Page(s) 32 in print issue
|
Up & Running
Tips To Improve Your Email Uptime
|
| Key Points
• Plan for downtime so that even if it’s unexpected, your plan matches the IT capabilities you have.
• Once you create a downtime plan, test it out to make sure it really works the way you want it to.
• Monitor your server’s health proactively and don’t make do with the least amount of capacity. | What’s the easiest way to go from hero to zero? Email downtime. At any company where employees can’t send email, communication falters, relationships with partners and customers suffer, and executives get into more of a panic the longer the downtime continues.
A recent survey by Neverfail (www.neverfailgroup.com), an Austin, Texas,-based messaging infrastructure company, polled 220 IT managers at midsized companies to ask them what the consequences of email downtime were. The results revealed that 25% of respondents said they would lose significant employee productivity, and 19% said they would lose revenue during email downtime. But there were other fears, as well: 34% said that it would endanger company or customer relationships, and 6% said that regulatory compliance would be affected.
Nearly three-quarters (74%) said that their company could only tolerate one hour or less of email downtime. Less than a quarter (24%) could tolerate four hours or more of email downtime, and none said that 12 hours of downtime was tolerable.
Obviously, every company is going to have email downtime now and again. It’s a matter of managing your planned downtime and minimizing your unplanned downtime. Here are some tips to help you do both.
 Hope Is Not A Plan
Start by thinking about how much downtime your company can take and formulate your plan for dealing with it accordingly. “Your contingency plan depends on an understanding of downtime’s impact,” says Andrew Barnes, senior vice president of corporate development for Neverfail. “If losing email for short periods would have little impact, then ordinary backing up to recover from a server crash is a solution. But more impact dictates a replication strategy so that you don’t have to take the time to go to backup tapes or disks to rebuild. And if downtime is going to have a critical impact, that means [you need] an availability option that doesn’t require a recovery process at all.”
Those different levels require an honest assessment of the IT skills you have onsite, as well. If your operations are open 24/7, do you have someone who is available 24/7 to address any failures, whether hardware, software, or both? Or, are you a nine-to-five shop that’s providing 24/7 availability to email users? Can you schedule server maintenance that will cause planned downtime for late in the evening, or will that involve paying overtime? If your company is trying to offer email availability that it can’t always deliver, a hosted email service might make more sense. But formulate your plan based on what you have, not on what you could have or think you should have.
Proactively Manage Server Health
The most frequent cause of email downtime is ordinary server downtime, whether that’s due to hardware failure or software configuration issues. A simple hard drive failure in the server will take the email down with the server. But so will running out of hard drive space or a power failure if you don’t have a backup power supply. “Tools that monitor disk status, detect an imminent crash, and monitor the health of your server’s configuration help you to keep your server—and your email—up and running,” says Barnes.
Avoid Resource Exhaustion
Messaging is a cost center, not a revenue generator, and that’s why many companies make do with the least amount of infrastructure. The key to avoiding unplanned downtime is redundancy, such as a backup server, so that even if one server goes down, the system can fail over to the other server and mail can continue to flow.
Bob Boucneau, senior consultant for professional services for Sendmail (www .sendmail.com/sm), a messaging infrastructure company in Emeryville, Calif., also suggests that any company that is without some kind of spam filter is asking for unplanned downtime. “Especially with older systems, if you’re not doing some kind of preprocessing for spam, mail volumes can suddenly go up by 10 or even 30%, and suddenly the server can’t handle the volume at peak times,” says Boucneau.
Follow Best Practices
Whatever your company’s resources, there are practices that minimize downtime. For example, any reconfiguration, patch, or other maintenance that takes the server down should happen after 5 p.m., preferably on a Friday—not at 10 a.m. on Monday. Patches that fix low-risk problems can wait until they’ve been tested, or at least until lower volumes of traffic are on the network. The firewall and the file server should never be on the same server. “Many companies have both on the same server, and they shouldn’t do it; it amounts to trying to protect your email server on your email server,” says David Setzer, CEO of Mailprotector (www.mailprotector.net), an email and Web security company in Greer, S.C. “Whatever you can do to move threats off the server and out of the network—for example, your Windows-based server should not communicate with the Internet at large—will help protect your email availability.”
Test Before Publishing
Software difficulties can take down a server, as well, particularly network issues such as an incorrectly configured database or router or a tangle between DNS, the firewall, and a router. “Networks are more complex than people realize, and when it comes to firewalls and routers, there are probably only about 10 people in the world who understand them completely,” says Boucneau. “They’re a big area of human error, and the mistakes people make can be hard to troubleshoot.” For that reason, if it’s an option, new configurations should always be tested before they’re visited upon the whole network.
But what if you can’t afford the costs of a testing environment? That’s when a hosted email service might make more sense, Setzer says. “Redundancy is expensive, and you can have the benefits of redundancy without bearing the sole cost,” he says.
Develop & Follow A Test Plan
No matter how carefully you monitor, the factor that most often causes downtime is ordinary human error. That’s why focusing on more exotic (if still possible) causes such as malware, viruses, or denial-of-service attacks can be detrimental. “Software today is so complex that human error causes a lot of problems, including email downtime, and the amount of training to know how to manage it is too cost-prohibitive for many companies,” Setzer says.
For that reason, a test plan to make sure that redundancies are working is essential. For example, if your plan is to send mail to an alternate server if the main server goes down, test to make sure that really happens. “Every organization should regularly power off the servers, singly and in groups, to test automatic failover so that if a real failure occurs, there will be no doubt that mail continues to flow,” Boucneau says. 
by Holly Dolezalek
More Email Uptime Tips
Avoid resource exhaustion. One way to ensure email uptime is to have one more server than is required to process mail.
Have alternate routes. Each server that processes mail should have multiple inbound/outbound routes to avoid backups or failures due to downstream network issues.
Consider a hot backup. “Many organizations use a ‘failover’ disaster recovery model where the disaster recovery data center is, essentially, cold until need arises,” says Bob Boucneau, senior consultant for professional services for Sendmail (www.sendmail.com/sm). “But for backbone infrastructure services like email, LDAP, or DNS, this is the wrong model. These services should be active and in use for both the primary and disaster recovery data centers to the greatest extent possible so that when a failover is initiated, these core services are absolutely certain to be available.“
Remember all your users. Email downtime affects not only onsite users, but users who access email over the Web, Windows Mobile, or their BlackBerrys.
Think globally when you act locally. If you take down the server for maintenance at midnight in California, remember that it’s 8 a.m. in London. |
|
|
