Chat iconGet in touch
Uncategorised

More Speed, Less Expense – Why Telcos need a next generation database

By April 8, 2017 No Comments

In addition, operators would like to capitalise for analytics purposes, the vast amount of data generated by networks.  So there are ever increasing demands on the systems operating and managing the network.  All of this means there is now significant evolutionary pressure on Service Providers to adapt to this new environment. This has been recognised by the operators themselves, who now seek to address it through such initiatives as the ETSI lead Network Functions Virtualization specification. This proposes to leverage computer virtualisation as a way to deliver elastic scalability, faster roll-out and reduced operating costs.

A key requirement for systems to enable this evolution is the ability to scale to handle massive TPS rates at low latencies. But this must be achieved in a way that still maintains the service levels and robustness demanded by Service Operators.

 

The way forward

One of the keys to this is the choice of database technology. The industry is now looking beyond traditional SAN backed RDBMS’ to modern in-memory databases that are faster, have better scalability and are less costly to deploy.

Openet conducted extensive trials and evaluated a large number of both Open Source and commercial offerings available today.  Both relational and NoSQL offerings were considered in order to select the “Best of Breed“ technology that met the needs of service providers. The majority of the products evaluated were fast and scalable, but few could continually meet the low latency and consistency requirements demanded for real-time systems.  Even only occasional “glitches” in such systems would have real world consequences for operators in terms of Regulatory, Revenue or Customer Experience issues.  Products that offered “Eventual Consistency”, required Disk storage (single point of contention) or Back End Storage and related I/O for current processing simply did not cut it in this environment.

 

Why VoltDb?

For the sake of brevity, I’ll refrain from an in-depth exposé of why we made our choice; essentially, VoltDb is the product that best fits the demanding needs of Openet’s customer’s real-time processing requirements. However, some technical highlights include:

  • An Ability to sustain Massive Throughput with Near Linear Scalability
  • Supports a fully Clustered system whilst maintaining ACIDity (Atomicity – Consistency – Isolation – Durability)
  • A VM friendly High Availability capability that uses simultaneous deterministic execution at multiple locations instead of streaming changes to a standby DB or a disk drive
  • The Shared Nothing model is a compelling design pattern for handling the ‘atomised traffic’ profile of Telco events
  • A comparatively small resource footprint for high TPS leading to lower TCO
  • Simpler Geo-Redundancy implementations and an ability for in-service upgrades minimises cost

Although VoltDB was selected purely on the its technical merits, it is reassuring that it is developed by a team with a significant Industry Pedigree including Michael Stonebaker, a founder, who has been a pioneer of data base research and technology for 20 years.