SSAS - Storage Options
More Details;
Real-Time ROLAP : All detail data and aggregates are queried directly from the
relational data source.No notification is necessary. No proactive caching is used. This may result in slow query performance, but data is always current.This setting is best for data that is changing frequently, leaving no time for cube processing, but which must always be up-to-date.
Real-Time HOLAP : Detail data remains in the relational data source. Aggregates are
in multidimensional storage. When Analysis Services is notified that the aggregates
are out-of-date, it processes the cube. It does not wait for a silence interval. While the
aggregates are out-of-date or being processed, queries are sent directly to the relational
data source. No proactive cache is used. This provides better performance for queries
when the aggregates are up-to-date, but reverts to slow performance while processing.
This setting is best for data that is also changing frequently, but provides some
intervals for processing.
Low-Latency MOLAP : Detail data and aggregates are in multidimensional storage.
When Analysis Services is notified that the aggregates are out-of-date, it waits for a
silence interval of ten seconds before beginning processing. It uses a silence override
interval of ten minutes. While the cube is processing, queries are sent to a proactive
cache. If processing takes longer than 30 minutes, the proactive cache is dropped and
queries are sent directly to the relational data source. This provides fast query response,
unless processing takes longer than 30 minutes. Maximum latency is 30 minutes.
This setting is best in situations where query performance is important but data must
remain fairly current.
Medium-Latency MOLAP : Detail data and aggregates are in multidimensional
storage. When Analysis Services is notified that the aggregates are out-of-date, it waits
for a silence interval of ten seconds before it starts processing. It uses a silence override
interval of ten minutes. While the cube is processing, queries are sent to a proactive
cache. If processing takes longer than four hours, the proactive cache is dropped and
queries are sent directly to the relational data source. This provides fast query response,
unless processing takes longer than four hours. Maximum latency is four hours.
This setting is best in situations where query performance is important and a bit
more latency can be tolerated.
Automatic MOLAP : Detail data and aggregates are in multidimensional storage.
When Analysis Services is notified that the aggregates are out-of-date, it waits for
a silence interval of ten seconds before it starts processing. It uses a silence override
interval of ten minutes. While the cube is processing, queries are sent to a proactive
cache. The proactive cache is not dropped, no matter how long processing takes. This
provides fast query response at all times, but it can lead to a large latency if processing
is long-running.This setting is best in situations where query performance is the most important
factor and a potentially large latency can be tolerated.
Scheduled MOLAP: Detail data and aggregates are in multidimensional storage.
Analysis Services does not receive notification of data source changes. Instead, it
processes automatically every 24 hours. While the cube is processing, queries are sent
to a proactive cache. The proactive cache is not dropped, no matter how long processing
takes. This provides fast query response at all times, but it has a maximum latency of
24 hours, plus the time required for processing.This setting is typically used in situations where a notification mechanism is not available or where data updates occur nightly.
- ROLAP (Relational Online Analytic Processing)
- MOLAP (Multidimensional Online Analytic Processing):
- HOLAP (Hybrid Online Analytic Processing):
More Details;
Real-Time ROLAP : All detail data and aggregates are queried directly from the
relational data source.No notification is necessary. No proactive caching is used. This may result in slow query performance, but data is always current.This setting is best for data that is changing frequently, leaving no time for cube processing, but which must always be up-to-date.
Real-Time HOLAP : Detail data remains in the relational data source. Aggregates are
in multidimensional storage. When Analysis Services is notified that the aggregates
are out-of-date, it processes the cube. It does not wait for a silence interval. While the
aggregates are out-of-date or being processed, queries are sent directly to the relational
data source. No proactive cache is used. This provides better performance for queries
when the aggregates are up-to-date, but reverts to slow performance while processing.
This setting is best for data that is also changing frequently, but provides some
intervals for processing.
Low-Latency MOLAP : Detail data and aggregates are in multidimensional storage.
When Analysis Services is notified that the aggregates are out-of-date, it waits for a
silence interval of ten seconds before beginning processing. It uses a silence override
interval of ten minutes. While the cube is processing, queries are sent to a proactive
cache. If processing takes longer than 30 minutes, the proactive cache is dropped and
queries are sent directly to the relational data source. This provides fast query response,
unless processing takes longer than 30 minutes. Maximum latency is 30 minutes.
This setting is best in situations where query performance is important but data must
remain fairly current.
Medium-Latency MOLAP : Detail data and aggregates are in multidimensional
storage. When Analysis Services is notified that the aggregates are out-of-date, it waits
for a silence interval of ten seconds before it starts processing. It uses a silence override
interval of ten minutes. While the cube is processing, queries are sent to a proactive
cache. If processing takes longer than four hours, the proactive cache is dropped and
queries are sent directly to the relational data source. This provides fast query response,
unless processing takes longer than four hours. Maximum latency is four hours.
This setting is best in situations where query performance is important and a bit
more latency can be tolerated.
Automatic MOLAP : Detail data and aggregates are in multidimensional storage.
When Analysis Services is notified that the aggregates are out-of-date, it waits for
a silence interval of ten seconds before it starts processing. It uses a silence override
interval of ten minutes. While the cube is processing, queries are sent to a proactive
cache. The proactive cache is not dropped, no matter how long processing takes. This
provides fast query response at all times, but it can lead to a large latency if processing
is long-running.This setting is best in situations where query performance is the most important
factor and a potentially large latency can be tolerated.
Scheduled MOLAP: Detail data and aggregates are in multidimensional storage.
Analysis Services does not receive notification of data source changes. Instead, it
processes automatically every 24 hours. While the cube is processing, queries are sent
to a proactive cache. The proactive cache is not dropped, no matter how long processing
takes. This provides fast query response at all times, but it has a maximum latency of
24 hours, plus the time required for processing.This setting is typically used in situations where a notification mechanism is not available or where data updates occur nightly.
No comments:
Post a Comment