One of the challenges for any IT department is to keep data available for those who need it. In fact, vendors and consultants have built whole businesses around the concept. Over the next few weeks, I'll be taking a look at a few methods for keeping data "alive" and available to end-users. Each method I'll cover is effective but depends on what you need to accomplish and what type of budget you have.
In a disaster (which could be anything from a failed power supply to an F5 Tornado taking out the building), the first thing you want to do is get the equipment running. But a very close second is making the recent version of data immediately available.
There are three main sources for replication: the host (servers), the array (specialized storage devices), or the network. Choosing the right method isn't always easy, but with some understanding of the differences, as well as your environment and needs, you can make a well-informed decision.
The first method we'll look at is host-based replication. Simply put, host-based replication is replicating data and using the servers that hold or receive the data to do the work of moving the data. Host-based replication has many advantages.
Advantages of host-based replication
- When you have a small number of servers that require data replication, you can tailor the host-based solution to meet your needs. Customization is key, since you can choose specifically what data to replicate between hosts.
- You can make purchases as needed and phase hosts into the replication strategy.
- You can pick and choose the data that is to be replicated (the entire server doesn't have to be replicated).
- You can create a schedule for sending data. Depending on the software you're using, you can throttle bandwidth based on availability and use (i.e., you can increase it in the evenings when the WAN use is less).
- Some solutions allow you to do a many-to-one sender/receiver replication.
- You can have dissimilar storage solutions at each end (Local storage to SAN or NAS, etc.).
Disadvantages of host-based replication
Here are a few disadvantages to using host-based replication:
- It requires a sender and receiver to be active on each end. This means that hardware and an OS (and possibly other applications) are required and increase your budget.
- Both the sender and receiver hosts must have the same patch levels for the OS and applications in order to do a true recovery in times of failure.
- With the numbers of servers replicating increases, so does their management time.
- Not all applications can support host-based replication.
Examples of replication software
Here are a couple of replication software packages that are currently available: