What Enterprises Need: Platforms that Support High Availability
By Jean Ouellette | June 1, 2011
2 minute read
This is the third installment in our weekly technical blog series about Enterprise Content Transformation: What Enterprises Need. The series kicked off with an overview of the reasons your organization needs an ECT platform, and last week discussed the importance of linear scalability when investing in business applications.
What is the cost to your organization when your critical business application is down? Some of our customers estimate the cost of not being able to transform documents at $100,000 an hour or more. Some examples are:
- A life sciences company cannot start its manufacturing production line on time because the PDF documents defining the production parameters are not available.
- An insurance company loses business to its competitors because they can’t produce insurance quotes PDFs when the system is down or has a slower response time than its competitors.
- A life sciences company must delay its submission to regulatory bodies (e.g. FDA) because the PDF documents are not all created. Keep in mind that one pharmaceutical product typically generates $1M of revenue per day.
It is because of critical business applications like these that Adlib supports high availability (HA). Let’s face it, organizations’ IT infrastructure are getting more and more complex, which can result in more points of failures. These range from physical servers, their electrical power and cooling systems, and the operating systems and applications that run on them.
Unscheduled system downtime can be minimized by eliminating all single points of failure. This means the application software must support an HA architecture. Adlib achieves this by supporting redundancy for all software components in the system. But these days, it is not desirable to have redundant node idle until a failover occurs because of the high costs (e.g. hardware, power, real-estate etc.). To eliminate this, Adlib implemented support for redundancy such that all nodes are active (called Active/Active) at all times and share the load. If one node fails, the system still provides service (at a reduced capacity).
Adlib components send heartbeats to the system database, and components that share the same function monitor each other’s heartbeat. When a component heartbeat stops, another component takes on its tasks within a few seconds. Alerts can also be configured to notify IT staff once a component failure occurs. This allows them to take action to bring the system up to full capacity and achieve the Service Level Agreement (SLA) committed to their users.
Mean time to repair (MTTR)is also an important factor to consider when deploying critical business systems, since they typically require 99.99% availability (resulting in only 52 minutes of allowable down time per year). Adlib has made recovery easier by storing all of its configuration data in the system database. This makes the provisioning of a new node as simple as installing the component software and defining the database connection for it.
Some of our customers tell us that users call IT for support if they don’t get their PDF within 3 minutes from making the request. Implementing a platform that provides High Availability and is easy to provision new servers will reduce support calls and keeps business running smoothly.