5 Secrets of Highly-Available Integration Infrastructure

Progress Software describes five misconceptions or "dirty little secrets' about high availability for IT integration infrastructure. The report discusses failover clusters and says five-nines uptime may not reflect true business costs of failures.

Tags: Integration Infrastructure, Transactions, Failure, High Availability, Transaction Processing, Vandervoort, Recover,


Progress Software describes five misconceptions or "dirty little secrets' about high availability for IT integration infrastructure. The report discusses failover clusters and says five-nines uptime may not reflect true business costs of failures.

Most integration infrastructure use clustering to achieve high-availability, recover from failure and restore services, according to Hub Vandervoort, CTO of Progress Software's SOA Infrastructure. "However, the bad news is that there is often a gap between what the business expects and what this infrastructure actually delivers."

Rather than measure high-availability in uptime, Vandervoort suggested it should be measured in terms of 'recovery time.' The rational is simple, Vandervoort said: Put simple, customers care about how long their systems will be unable to process transactions during a failure.

According to Progress Software, the five dirty little secrets of highly available integration infrastructure are:

  • The Myth of Five Nines. Traditional measures of availability don't track what an enterprise's IT customers care about—how long it will take to process their transactions when systems fail. Recovery time. Conventional, highly available integration infrastructure actually interrupts transaction processing for 5-15 minutes when they recover from failure.
  • Data corruption. Conventional, highly available integration infrastructure may deliver transactions twice or in the wrong order when they recover from failure.
  • Scheduled maintenance. Conventional, highly available integration infrastructure needs to be brought down in order to upgrade hardware or software.
  • Hidden cost and complexity. Conventional, highly available integration infrastructure requires third-party products, application modifications and complicated operational procedures.

  • In summary, the Progress report concludes high availability through clustering doesn't deliver what business managers are looking for - always on availability

    "Unless an integration infrastructure provides 'continuous availability'—meaning that transaction processing continues uninterrupted despite system failure—infrastructure called 'highly available' is actually only 'mostly available,' Vandervoort said.



    back