I'm new to SAS Visual Analytics and I'm asking me, what exactly is the SAS LASR Analytic Server? I have read a lot of websites and for me, it looks like, that the SAS LASR Analytic Server is a component that I can use to load data from source tables into in RAM. But where is my data stored when I load a CSV table into Visual Analytics Source Table? Thing is, I come from QlikView and there is all the data always in memory.
Regards
LASR is SAS's previous-gen in-memory analytics engine that can either be distributed or single-server. When you load a CSV into LASR, it is converted into a special format and stored on the LASR analytic server. If you have a distributed setup, your data is distributed in-memory among your worker nodes.
You can access your data via the SASIOLA engine and a data step and use it just like any other SAS engine. High Performance procedures such as PROC HPSUMMARY, PROC HPLOGISTIC, and PROC HPDS2will take advantage of tables in LASR and perform very fast calculations on them. PROC IMSTAT can create statistical models in-memory, among other things.
You cannot overwrite LASR tables: they must first be dropped and then reloaded. You can, however, append, delete, and update rows using PROC IMSTAT; in my experience, it's just easier to drop and reload tables unless you have very specialized report uptime requirements.
Tables cannot be accessed from Visual Analytics without first registering it to Metadata. VA will automatically handle this for you if you load data directly.
Your data will stay in-memory within LASR until the server reboots or the table is dropped. Unlike Cloud Analytic Services (CAS) in Viya, LASR does not do automatic on-demand loading. Instead, you can set up an Autoload directory that will automatically update and refresh tables in LASR. SAS tables, Excel, and .csv files you drop there will automatically be picked up and loaded every 15 minutes. This is a good way to ensure your data is available after reboots, as it will scan this directory and load data.
In general, my process has been this:
If you are lucky enough to have co-located Hadoop, you can store your data as a sashdat file and load data extremely quickly into LASR. In essence ,sashdat is LASR format on-disk. Data can be automatically loaded into memory on reboot from there, but it will not behave the same as Autoload.
Cloud Analytic Services (CAS) has superseded LASR in recent years and builds heavily off of its existing features.