Spending countless hours and millions of dollars on IP service testing, only to get unacceptable quality?
Network operators, product vendors, government agencies, smart grid and the cloud computing ecosystems are all struggling to deliver IP Services to the market with high quality, on schedule and within budget. Test organizations are being asked to do more with less resources and it’s clear that current approaches to testing and developing IP Services is not sufficient.
Existing testing approaches are characterized by excessive costs,
time-to market delays, unacceptable quality, and deployed services.
According to analyst researcher NSP, the following industry trends are
commonly found:
- The average time to test a new IP service is more than 12 months
- The average cost of testing a major product release exceeds $4 million, or 10% of product development costs
- The average test coverage on a first release is less than 40%
Manual testing is highly labor-intensive: Testers write code and develop test scripts for their specific needs. The sheer number of combinations of inputs, configurations and versions that need to be tested makes this approach scale poorly over time.
Network load testing tools are commonly found in the tester’s toolkit, but they are first and foremost designed to be infrastructure-level load generators or “packet blasters.” They
offer a set of static pre-built tests based on standards-compliant traffic, which means they
don’t easily test proprietary extensions or non-standard protocols. Their primary function
is to ensure that a device sustains production-level network throughputs, generally for
one protocol at a time.
Since IP Services depend on a sequence of multiple protocols working together, single protocol load testing offers minimal test coverage. Also, since the traffic in the network is often based on custom extensions, custom network configurations and variations from
the defined standard, this approach does not accurately represent real-world situations.
The legacy approach to IP Service testing is clearly not sufficient for your sophisticated systems. It’s time to take a fresh approach!
|
