Over the past couple of months we’ve been focusing our development efforts on improving our hosting and associated product environment via an appropriately titled ‘production infrastructure sprint’.
Although this doesn’t sound as exciting as adding features to our products, it’s a vital part of Delib’s service to our customers, as it helps to ensure that we continue to meet our uptime and performance commitments. Here’s a little overview of what we’ve been up to.
What we’ve been doing
Up until recently we hosted all our customer instances on large multi-tenancy servers. ‘Multi-tenancy’ means that several Delib customer sites run side-by-side on the same machines, although all their data is stored in separate databases. These servers live in secure data centres, physically located in the same territory as the customers they serve. The data centres are responsible for providing Internet connectivity for the production servers.
Over the past few weeks, we’ve been moving customers slowly and carefully in batches from our current hosting providers to new providers who can better meet our service and uptime requirements.
Why we’re changing our hosting infrastructure
The reasons for migrating to new hosting providers are threefold:
1. Improvements in availability
In the UK, we are moving all our hosting to Rackspace, the market leader in cloud hosting, which offers a 100% uptime guarantee. Since our uptime is necessarily bounded by that of our upstream providers, it’s important to use the best that we can get. We are researching the best providers in other territories, to ensure that we continue to meet and exceed our commitments for all our customers.
We use a server monitoring service that notifies our account managers and developers by text message whenever a customer’s instance is unavailable for any reason (even if it’s in the middle of the night) so we’re all keen to ensure that these improvements pay off as soon as possible!
2. More hosting options for customers
After migration, every Citizen Space and Budget Simulator instance will live on its own virtual machine. This allows us to offer different hosting packages for different usage patterns: we can now tailor the system specification (RAM, disk space, number of processors) to the requirements of the customer. Furthermore, large spikes in one customer’s traffic can no longer adversely affect the response times of other customers’ sites.
Dialogue App instances will continue to run on a multi-tenancy setup by default. However, customers with heavy usage requirements (eg large, heavily-publicised national dialogues), will have the option to host their Dialogue App instance on its own machine.
3. Consistent configurations and automation
As our number of customers grows, our developers have been spending more and more time engaged in administrative tasks such as rolling out new instances and upgrading existing customers. While this is vital to the business and to our customers, developers would much prefer to spend their time developing new features and fixing bugs in the products.
At the same time as moving customers to the new hosting infrastructure, we’ve been improving our suite of developer tools so that more of the day-to-day tasks can be done without developer intervention.
For our customers, this means that planned maintenance should soon be able to take place, as far as possible, outside working hours. It also means that developers will have more time to spend on improving our products, resulting in a better user experience for our customers and end users.
Find out more
If you are interested in finding out more about the improvements we are making please feel free to get in touch with either Louise or Rowena.