The problem

Scalability defines a system capacity to reply linearly to a growing request load.

As machines have limits (computation, memory limits), to ensure a good scalability, one must make sure at any moment that there are enough available servers to serve all requests; otherwise:

• at best, system response times are greatly increased,
• at worst, some requests are denied,
• or even worse, servers will die of memory shortage (especially if applications are badly conceived). A manual intervention by the IT administrator will be required to restart them.

High availability is expressed in percents by year. For instance, if a system is available at 99.95%, it means that it will be unavailable less than 0.05% per year, that is less than 5 hours per year. One can consider this figure as high availability, since, in real life, servers are usually unavailable several times a month or a week (because of software upgrades, hardware maintenance, deployment issues etc.).

Thus, scalability and high availability of your servers are critical factors for the success of your enterprise:

• at a lower scale, inside your very enterprise, if you add many client PCs, you will need to replicate your servers, or they will not be able to answer efficiently to all requests. You are risking unavailability or service discontinuity and with it, a productivity loss of your employees.
• at a medium scale, if you execute from time to time massive computation through mathematical software (or 3D or CG software etc.), you will need to replicate your servers, or your computations will take too long to complete. You are risking a big expense in new servers, hard to amortize for this occasional need.
• at a bigger scale, if you are into e-business (you offer online sales), a great number of Internet clients trying to connect will overload or even crash your servers. You are risking again service discontinuity, and with it, huge loss of profit (your clients will leave your site if it is unavailable or simply too slow).

Before Cloud Computing, managing the scalability and high availability of your servers (Web or calculus) was slow, complex and costful.

The solution

Deploy your servers into the Cloud.

Promises and key arguments of Cloud Computing are precisely scalability and high availability.

Select the best Cloud Computing provider for your needs. Indeed:

• All providers take care of hardware maintenance of their datacenters; it is a big relief for your IT. But only certain providers take care automatically of software upgrades.
• Regarding deployments, some providers offer tools to test your applications in a preproduction environment identical to the production one. Once the preprod environment is validated, switching a production mode is instantaneous, thereby avoiding many downtimes and service discontinuity.
• Some providers offer an option at a extra cost for automated control software which supervise your servers life. Your servers are automatically restarted in case of crash.
  • This control software are more or less full of options.
  • Some of them also allow to monitor the amount of inbound requests so that new servers are started on demand. Thus you solve easily scalability issues.
  • In all cases, your Web administrator keeps a manual control over your servers, as depicted above.
• Finally, most of providers own geolocalized datacenters, in order to replicate your servers as close as possible to your users: response times are then a lot better, and contents can be personalized depending on location.

Contact us to learn more.