RECAP Validation Results Released

 

After three years of research, the RECAP consortium released results validating their proposed tools and methods.  We have been ambitious in our approach to the automation of resource management across a wide range of network and cloud use cases but the results for each case are very promising.

Some of the highlights include:

vCDN Infrastructure Planning and Management

The RECAP infrastructure simulation and optimisation method suggests that the choice of sites for CDN infrastructure in the UK network can have an impact of up to 24% on the network cost savings achieved by the caches.

The RECAP traffic forecasting method, when tested with real CDN statistics, could be used to reduce typical resource underutilisation from >50% to ~12% facilitating significant power savings.

RECAP also provides methods to optimise the amount of storage required for CDN caches. Simulations demonstrated clear cost minimums could be found saving up to 25% on compute and storage investments. These savings equate to several tens of millions of Euros in the case of a company such as BT.

LTE Infrastructure and Network Management

Our research suggests that RECAP autoscaling can save up to 24% of the resources without any compromise of Quality of Service (QoS) for the LTE Evolved Packet Core. RECAP’s method to optimise the placement of LTE network components demonstrated an optimal placement could out perform a sub-optimal placement using twice as many resources.

Resource Placement for Edge/Fog Computing for Smart Cities

RECAP methods demonstrated placing IoT processing data capability at optimum locations in a city could save up to 17% of network traffic compared to using a centralised Cloud service for its application of real time route calculation for vehicles used data from the city of Cologne.

Infrastructure Optimisation for Cloud-based Big Data Analytics Services

RECAP’s worked with Linknovate, an SME cloud-based Big Data analytics service. The simulation model helped us understand how the ElasticSearch search engine works. By modelling different scenarios where are able to evaluate how the CPU and RAM usage vary based on the number of data nodes responding to a query. This can help Linknovate understand their system architecture and how they can improve/scale it in order to support their desirable query traffic. Adequate application dimensioning helps to manage costs of cloud services.

Our suggested changes could result in an operational cost saving up to 80%.

Download the Deliverable

Our report on the validation results, D3.3, can be downloaded here.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.