The Cloud: Focusing On Cloud Performance – Part 1
The cloud has established its presence with applications for individuals, groups, and small and medium businesses. As a result, independent software vendors who have peddled packaged applications to cater to business functions in different verticals for companies of size might be pondering over when it is right to transfer some key offerings to the cloud and transform themselves into cloud service vendors. Unlike small players who can leverage cloud platforms like Amazon or Google Apps, they may need to create their own or shared platforms to provide their services. This is because the workloads they will have to cater may not be amenable in the existing platforms for many reasons, magnitude being the foremost.
While in the cloud a lot of caution and thought is put into addressing issues related to security, standardization, and regulation, performance is an area where excitement seems to prevail. After all, features like availability, elasticity, auto-scaling, proactive scaling, and parallelization are supposed to solve performance problems and enhance capabilities. While this could be true for the end-user and client perspectives, these are performance requirements that have to be delivered and met from a vendor perspective. Catering to the business workloads of several sizable companies is in itself a huge performance feat for a potential cloud service provider. Add the above features and we are looking at a very impressive challenge to deliver in terms of performance. This message should not be lost on the community before attempting to take the cloud to the next level—delivering industrial functionality.
Traditionally, performance has been a nice-to-have feature. With packaged software meant to be deployed for individual companies, sloppy performance has often been an annoyance but rarely a showstopper. Moreover, issues related to performance have been localized to individual companies and their customers. Scouring through ‘Schadenfreude’ (as one author describes the websites by utilizing a German-based word that hints at pleasure felt at another person’s misfortune) containing websites that list software disasters will reveal only a few attributable to performance issues but which have had serious impacts.
In the cloud performance is critical. Performance issues could bring down industrial sectors and regions and could impact economies. In the cloud, not performing is not working. At the least, poor performance can eat into a CSV’s bottom line. There are two aspects where understanding size quantitatively matters immensely. One is the load that must be handled at the data center where the application providing the service is hosted. The other is network traffic that reaches into and out of the data center. This macro-level understanding is a must before building platforms or making service offerings for delivering industrial functionality through the cloud. Cutting corners to achieve deadlines or go to market first strategies or compulsions not to be left behind could hamper putting enough attention or caution here—that could be disastrous…
Part 2 – Next Week
By Suri Chitti
Suri Chitti (suri_chitti@hotmail.com) is a software and technology management professional. He has packed many years of experience and insight having worked in different industrial domains, for different stakeholders like companies, vendors and consultancies and has contributed to different aspects of software production like functional analysis, testing, design and management.
