Hosting standards and SLAs of eCommerce platforms

Picture of Piotr Karwatka
Piotr Karwatka

Former CTO of Divante eCommerce Technology Company. Open-source enthusiast and life-long builder. Co-founder of Vue Storefront and Open Loyalty. Now gathering engaged communities around new technologies.

Care to share?

Divante provides not only the software services. We also keeps our customers websites up and running on our hosting with high SLA. If you are interested how we do this please read this post!

Hosting standards, SLA guarantees

As part of the hosting service we provide not only the equipment and Internet connections, but above all, we ensure the continuity of operations and SLA parameters.

We guarantee fault repair times and their maximum durations per month. The contracts define the appropriate punishment when exceeding the parameters.

The hosting covers the following steps in order to ensure the highest-possible quality:

Infrastructure Project

Infrastructure Project

As a part of the analytical stage we prepare the infrastructure design (see the attached example). The design is based on business objectives – that is, the maximum number of viewers, transactions per unit of time, etc. The design is dependent on the software and requirements. It includes one or more servers, and describes their exact specification.

At this stage, we are building architecture solutions, choose the appropriate system solutions. Due to designing the infrastructure for specific requirements, there are no problems with its implementation at a later stage. The whole system architecture will be compatible with the technology concept.

Recovery Plan (speed, scope, schedule, procedures)

The Recovery Plan is contained in the documentation for administrators and describes all the stages to be taken to start-up again in case of failure. It describes how to grant authorizations, and install the application code, and which configurative files should be modified and how to obtain and restore the database backup.

Thanks to the plan – even at night, the people who have no experience with the system are able to take the minimum necessary steps to ensure the continuity of the software.

Security of data (including personal data)

We attach great importance to the security of information. All employees of Divante are trained in terms of security of personal data and sign a declaration of confidentiality and respect inside security policy.  Divante has a policy prepared by specialists and safety instructions – for hosting systems and for administrative staff. It is possible to have an insight into the extract of these policies. All servers of Divante are properly protected, overseen and audited, and meet the requirements of the Inspector General of Personal Data Protection.

High Availability (HA), scalability, redundancy

Technological concept – if required –  includes relevant high availability techniques. Normally we work on a model in which the servers are started-up within the virtual VPS hosts on the dedicated Divante servers.  In the event of server failure they can all be transferred to other efficient machine. Every few hours all the server state is backed up to another data center. In real-time mode databases are synchronized (replication). Due to this approach – in an emergency – we are able to quickly switch on the application into the emergency environment and to maintain its operation. In case of bigger requirements as to the availability, it is just the Technological Concept which describes the details and architecture of the solution.

Documentation for Administrators, procedures and QA lists

As part of the implementation and implementation Best Practices technical documentation for the system administrators is prepared.  Documentation helps take care of the system is a tutorial created by developers for administrators. It describes all the administrative tasks (e.g. regular cleaning of the database, logs, etc.) necessary for stable operation. The documentation also includes the response to frequently asked questions and complete solutions of the previously occurring administrative problems.

Full automatic monitoring and automation of tasks

We provide full monitoring of servers. We monitor not only the load of the server (RAM, CPU, disk …), but also operation of the application. The standard is that after the stage of development during the tests  automatic checks – i.e., scripts testing the basic mechanisms of software functioning are created. Every few minutes it checks the ability to viewing the system, adding products to the cart, etc.

We use ZABBIX and phpUnit systems. In the case of any problems the administrative staff is notified automatically and may take steps before the users notice a problem.

Redundancy of the administrative staff

We make several employees dealing with maintenance to have the knowledge of the system. In the case of holidays, illness or too much workload of one of them – the others will be able to take the action.

Each month, we provide SLA reports of behavior and information about the escalations- such as regarding security problems.

Microservices Architecture for eCommerce eBook. Download for free >

Published August 1, 2014