Big Data in HTML5

Backend Optimisation Options in SAS

Principles

• Separation of front / backend
• Presummarisation
• Paging
• Intelligent caching

Front vs Back

Separation of front and backend is non-negotiable!

• Scalability
• Forward compatibility (SAS 9 -> Viya)
• Speed of development (workflow)
• Differing skillsets
• Documentation of API
• Maintainability

Presummarisation

Send only the data you need to send!

• Browsers often have restricted memory on corporate devices
• SAS is fast and extremely powerful for data management
• Well defined interfaces minimise the need for client side rendering and large IO volumes
• Avoid 'stateful' services (eg STP sessions) as they can impact load balancing

Paging

Avoid browser-freeze!  When showing tables of data, divide into 'pages' that can be stepped over.

• To keep browsers responsive, they shouldn't be dealing with more than 1-2 mb of data at a time 
• Prefetching nearby pages will further improve responsiveness
• Requires STATE - this can be in a filesystem or database but involves cleanup

Caching

Browsers automatically cache JS / CSS / Image files etc.  Server side data caching significantly improves the responsiveness for large queries - simply store under a hash of the input query and use that as a lookup.

• Requires a cache management system (lookup tables)
• Proactive cache generation avoids the 'initial query' cost 
• Be sure to remove stale caches when source data is refreshed