Understanding the importance of data protection of messaging systems and its continuous availability is critical. Messaging today has become much more than just another application. “Most businesses are just beginning to realize that messaging has evolved from an application to a corporate communication platform,” believes Manish Kalia, founder and vice president of marketing for Teneros, Inc.“Email, BlackBerry/iPhone, voicemail (unified messaging), fax, archiving and compliance, document management and CRM are just a few of the IT systems that are dependent on messaging. As a result, messaging has emerged as the number one application that needs to be protected for disaster recovery and high availability.”
That dependency upon messaging needs to be considered, when addressing disaster recovery (DR) and high availability (HA). “When it comes to protecting a messaging system, one of the things we are keen on talking to customers about is whether the organization is protecting messaging or other components of the IT infrastructure,” says Eric Pitcher, VP technology strategy for the recovery management and data modeling business unit at CA. “You cannot have DR just for an individual component. If you only worry about DR for backup, then getting back the individual file is easy. However, getting back the full email server is not. One of the things about applications—and email is just as susceptible as any other application—is that it lives in an IT ecosystem. It has to have other services available in order for email to work and be accessible, such as security, your LDAP, or Active Directory or whatever you are using. Making sure your messaging systems are seamless and how you protect your IT infrastructure is critical.”
Andrew Barnes, senior vice president of corporate development for Neverfail Group also believes that messaging has extended its reach, “When people are looking at business continuity it is no longer an application in isolation. It is no longer just about Exchange, but about everything else that goes around it.” Barnes then asks: “Do you want three or four different tools to manage three or four different types of applications? Or do you want one product or one interface to manage it? This is a good debate to have. How many tools do you use to deliver HA and DR protection of messaging and messaging related systems?”
Virtualization and Business Continuity
Another consideration is the complexity of the infrastructure. Is it a physical or virtual environment? Often today’s IT ecosystem is mixed. Where does the virtualization trend enter into HA and DR? “One thing we have been hearing a lot from the virtualization vendors is that the primary reason to buy virtualization is to reduce cost,” responds Pitcher. “This is done, of course, by allowing multiple applications to run on a single piece of hardware, but there are pros and cons to that.” He explains that while virtualization does help to reduce costs, it also adds complexity to the IT infrastructure. “If you are running a physical environment, you can have your DR environment be virtual,” he says. Pitcher notes that CA XOsoft takes care of all the complexities of replicating and creating application HA between a physical and virtual environment, “As a matter of fact it is the most common install I see of our technology today: messaging systems running physical, but their DR site is running virtual with CA XOsoft protecting that environment.”
Virtualization, while bringing benefits to business continuity, does require careful planning. Barnes reminds that while virtualization can be a very reliable stable platform protecting against hardware failures, it may not recognize what is going on within an application. “It is important to look for ways in which you can understand what is going on inside the virtual machine,” he cautions. “Neverfail can do that because it is application aware and doesn’t care if its running on virtual or physical, then you can make decisions about failover, based on what’s going on inside the application, inside the virtual machine.”
Kalia also sees the role that virtualization can play, saying: “Virtualization provides protection from hardware failure. However, that only accounts for a small percentage of failure scenarios in the messaging world. Common messaging failure reasons include: database corruption, software errors, storage failure or operator error.” Kalia recommends virtualization in conjunction with transactional database replication and application level failover/failback logic to provide a comprehensive DR solution for the messaging infrastructure. “For applications other than messaging, that are not databases, such as Web servers or file servers, virtualization can provide an effective DR solution by itself,” he adds.
With the increase of interest in virtualization, Pitcher notes that DR and HA implementations are aiding migrations from physical environments to virtual. “One of our number one customers actually bought our replication capabilities specifically for doing physical to virtual migration. One of the problems you run into with a virtualized environment is when you virtualize a server, you have to do the physical to virtual conversion—it’s usually all or nothing.” Pitcher states that, “A lot of CA customer sites are setting up a virtualized server, get it tested and running, ensure the impact that virtualization has on their DR plan is understood and then use CA XOsoft to replicate the email messaging system from a physical environment to a virtual environment. When they are comfortable with the DR environment running they do a manual failover, and shut down the physical server and bring up the virtual server. What’s neat about this is a customer has now migrated its application to a virtual environment, but if the virtual server is not performing, it still has the physical server available. It gives them an immediate backup, so that if anything fails, all they do is failback and everything is up and running again.”
Most experts agree messaging infrastructures do require special planning and consideration. “Messaging is not just about idle communication, it is often part of a critical workflow process, so what ever you use to protect email to keep it up and running you might want to consider if that’s the right technology to use for ancillary components, or vice versa,” suggests Barnes. He goes on to offer an example of a customer that had HA of Exchange and HA for BlackBerry, but with two different tools. “What if one of them goes down and the other doesn’t? You have a problem, because for the one that goes down you might want to switch somewhere else.” He continues by noting that the solution for this customer was Neverfail for RIM BlackBerry protecting BlackBerry service through a disaster site some 40 to 50 miles away. “But then what you get is potential latency because you get lots of network traffic where BlackBerry has to talk to Exchange to deliver email,” Barnes explains. “So they wanted their Exchange systems failed over as well, so they bought a product from us called Neverfail ClusterProtector, ensuring its Microsoft Exchange environment was able to failover to the same location as the BlackBerry Enterprise Server, so now both are in the same place and both available to the end-user. This is a good example of thinking about not just protecting one application, but protecting multiple applications and making sure that you maintain the optimum configuration when that happens.”
Planning, however, is not just about technical implementation. Pitcher notes that he works with organizations like Contingency Planning Exchange (CPE) where companies help each other with business continuity needs. “With business continuity planning only one-third of the problem is technical,” he believes. “Two-thirds of the problem is people and processes.” Pitcher notes that CA offers a booklet on its site: CA Business Continuity Planning: IT Survival Guide.
“The primary theme of the booklet is how critical messaging is to business continuity,” he reveals. “Since it’s the one piece of infrastructure that everybody relies on, the most critical thing is getting that communication back up and running, so that management can talk to the employees and customers and let them know what to do. In the primer, we give examples of the importance of and ways of keeping communications open.”
Barnes points out that DR can give a false sense of security. “To many people DR is sending a copy of the data offsite, or replicate offsite,” he says. “But if all you do is replication, you still have to recover a working system. You still have to make sure you have the right copy of Exchange, that you have a server available to run Exchange, and that Exchange is pointing to the database. That process can take a couple of hours, which is not usual if you have to rebuild the application server. That is two-hours without email and for some organizations it doesn’t matter, but for others it can be absolutely critical. So people will think that replication, and therefore data offsite, is DR and it is up to a point, but it’s not the same as disaster availability.”
With messaging a vital part of today’s business processes, many organizations realize that messaging infrastructure needs to be treated differently when it comes to HA and DR. “Messaging is a very database-intensive application as compared to other IT applications,” observes Kalia. “This requires special HA and DR planning through choosing the right technologies for messaging DR that is resilient to database corruption and data loss scenarios due to database divergence that is very common in DR.” As messaging is the glue for many IT applications, Kalia concludes: “As a result, if messaging is down, the entire IT infrastructure is affected and there is no communication collaboration capability left within the business.”
For recommendations for choosing a high availability/disaster recovery solution read:Preparing for a Failure Event.