Infrastructure – Data Centre for Virtual Contact Center Platform

Data Centres

Our infrastructure is located across a range of European data centres, all capable of operating independently of each other if required. Our primary data centre supporting UK and Ireland customers is based at DC Netwise (Netwise House, 24 Old Jamaica Road, London SE16 4AW). This is interconnected with the three other Datacentres in the EU. Two of these are based in Prague in the Czech Republic (DC Nagano U nákladového nádraží 3153/8, 13000 Prague 3 and TTC DC1 Tiskařská 257/10, 108 00 Prague 10) with a third data centre based in Slovakia (SWAN DC Údernícka 14, 851 01 Bratislava, Údernícka-Kopčianska).

The individual racks are in air-conditioned halls, with a high standard of safety, back-up internet connectivity and back-up power with batteries and diesel generators as a matter of course. In each of these data centres, we have prepared sufficient hardware resources in case of a long-term outage of one of the locations. All work and access are reported in advance to our administrators. If there is a possibility that a planned intervention in the server room will affect the availability of services, we publish a notice on status.daktela.com with a detailed description of the intervention.

DC Nagano (U nákladového nádraží 3153/8, 13000 Prague 3)

TTC DC1 (Tiskařská 257/10, 108 00 Prague 10)

SWAN DC (Údernícka 14, 851 01 Bratislava, areál Údernícka-Kopčianska)

DC Netwise (Netwise House, 24 Old Jamaica Road, London SE16 4AW, United Kingdom)

Network Infrastructure

Our datacenters are inter-connected by redundant high-speed fiber optic L2 lines. The fast connection provides sufficient capacity for online migration of virtual servers across sites as well as for daily backups. Our customers can access Daktela using their regular internet connectivity or using dedicated private line to our datacenter based on MPLS technology. The network topology is schematically indicated in the figure.

Hardware

We operate exchanges on two types of servers:

• Standard 1U Dell servers, which are ideal for us as they utilise the powerful and highly clocked CPUs of the Intel E3-12xx series (3.8GHz).
• We also use Supermicro servers of the MicroBlade series, which excel in their compactness and modularity. We currently have one box in each data centre with plans to add more as the requirement grows. The advantage of MicroBlades is compactness, where case 6U contains 28 blades on which up to 56 separate servers can be placed together. Another advantage is the very low power consumption while maintaining peak performance, redundant power supply, redundant switch and completely wireless connection of individual blades.

For hourly backups of exchanges, we use two powerful disk arrays with a capacity of hundreds of TB, consisting only of SSD disks.

Monitoring

All active devices located in our data centre such as: physical servers, individual virtual exchanges, switches, routers, UPS and GSM gateways are constantly monitored. We use the Zabbix monitoring system with a number of our own monitoring enhancements. We currently monitor approximately 1,000 guests, from which we regularly obtain 90,000 different values, over which we perform 32,000 automated events. In case of failures on important elements in Daktela Cloud, the system automatically calls and sends SMS to selected administrators. Quick resolution of critical issues is enabled by our support available 24 hours a day.

We also provide access to the monitoring for our clients if required.

Backup and restore

We perform backups of customer systems continuously in the 24/7 mode, where each backup is not more than 6 hours older than the previous one. Backups are alternately stored in two independent repositories in separate locations, so even if one DC is not available, no backup is older than 6 hours. We then move long-term archived backups to high-capacity ZFS storage.

Restoring a selected backup typically takes less than 10 minutes (depending on size).

Scenarios are prepared and regularly tested for critical situations of various kinds, from HW failures to power outages in our DC, which minimise any system downtime.