[Episode 03] One year after our network infrastructure upgrade

One year ago, I published on this blog an article on the upgrade of Pentalog’s WAN infrastructure. At the time, the project called WAN2009 was meant to allow us to:- Increase our bandwidth- Increase the quality of our services- Provide access redundancy to our internal resources and the InternetAll the resources envisaged were deployed: two level 2 links of 50Mbps between the data centre in Bucharest and the Romanian offices (2x20Mbps for the Moldovan office), 30Mbps Internet link provided by several operators for all its offices, level 2 link of 100Mbps between data centres in France and Romania.One year later, I would like to make a quick review in order to share the benefits and difficulties with you.Based on a few precise figures (estimated in certain cases), the average benefits for the Romanian and Moldovan offices are:- We have gone from an estimated 92% Internet access availability to 99.23% (measured between March and November 2010), i.e. a 10 times reduction of unavailability (from 60 hours to 6 hours).The remaining unavailability is 80% due to power cuts.- We have gone from an estimated 91% availability of access to internal resources to 99.1% (measured between March and November 2010)- The budget for the services provided has been doubled, while the bandwidth has increased 12 times for a three-month investment for previous services provided.- The supervision of this new infrastructure was carried out with no additional human resources.- We estimate a bandwidth saving of between 10 and 15% through the “simple” implementation of anti-spam and Web traffic filtering rules- We have estimated an eight-month ROI through availability gains and increase in bandwidth.On a practical level, the two office moves carried out in Romania since the deployment of WAN2009 have allowed us to greatly reduce the IT risk. Indeed, going to level 2 has enabled us to eliminate the entire IP address reconfiguration phase on network equipment (France + offices). Internet access migration has been carried out without any problems. The issues related to the synchronization of changes that operators repeatedly faced have now disappeared.In order to optimize our budget, we are managing the increase in Internet bandwidth in a single point where costs are the lowest and where we have several accesses, for redundancy and competition reasons.By deploying identical equipment in all points which are easier to manage, we have avoided the integration of a new high-level profile (VPN, redundancy…). This was done by using competences that were already in place. This centralized management also implies an improvement in the security level through practice convergence. This equipment has also enabled us to implement traffic prioritization rules in order to favour voice communication, internal and customer communication and by separating professional traffic from other types of traffic (YouTube, Google etc.). In the end, we obtain a better service quality.Without this infrastructure, now we wouldn’t have the same level of service quality on the Centrex that we now manage on our own. But there are still improvements to be made in order to increase the quality of our services.This star network between Romanian offices around a data centre can be reproduced in Vietnam when we open several offices there. We have already identified the different players but the best way to benefit from this architecture is with two or three offices.The actions we had planned for 2010 were:- The virtualization of work stations (VDI): Experiments have not progressed enough so that we can generalize. But the infrastructure that we are preparing will allow to meet our needs. On the other hand, we must still ensure the benefits of this approach. Projects like ProductionFabric will inevitably need this type of infrastructure.- The study on deploying a video-conferencing solution is underway as part of a cost reduction policy.Besides these projects and the deployment of our cloud computing ecosystem, in 2011 we intend to redesign our backup system and implement the disaster recovery plan for our entire activity.[Episode 02] How to avoid ignoring our HEROs[Episode 04] Thoughts on the VDI