Learn how to create, modify and run Firebolt engines.
S
:
M
:
MIN_CLUSTERS
and MAX_CLUSTERS
parameters to enable auto-scaling and allow the engine to adjust the number of clusters based on workload. Firebolt scales the clusters between the defined minimum and maximum based on engine CPU usage, time in the queue, and other factors that vary with demand. Auto-scaling helps your engine adapt to fluctuating workloads, improving performance, minimizing delays during high demand, avoiding bottlenecks, ensuring consistent query response times, and optimizing resource utilization for a more cost-effective solution.
For a technical guide on how concurrency auto-scaling works, see the Understanding Autoscaling page.
To use auto-scale, do the following:
MIN_CLUSTERS
set to a value and MAX_CLUSTERS
set to a value higher than MIN_CLUSTERS
as shown in the following code example:
MIN_CLUSTERS
has the same value as MAX_CLUSTERS
, auto-scaling is not enabled.
information_schema.engines
view to check how many clusters are being used by your engine. The following code example returns the number of CLUSTERS
, MIN_CLUSTERS
, and MAX_CLUSTERS
from the specified engine:
information_schema.engines
to observe the change in the CLUSTERS
value. You can use any query to test this functionality as long as it can overload the engine. The following example is one such query, but you can use any query that causes the engine to overload.
a.x
and b.y
after casting them to BIGINT
, and the total count of joined rows from two generated series of numbers ranging from 1 to 1,000,000:
CLUSTERS
as 2
as shown in the following table:
clusters | min_clusters | max_clusters |
---|---|---|
2 | 1 | 2 |
INITIALLY_STOPPED
function can only be set during engine creation and cannot be modified afterward.