Falarica Query Platform is a multi cluster, analytical workload manager. It supports standard ANSI sql. It uses Falarica’s Presto Operator to deploy one or more Presto clusters, on one or more Kubernetes clusters which can be on the same cloud, on-premise, multiple different clouds and zones or their combinations. It provides a single pane of glass to the end users to fire sql queries and exempts them from bothering about the logistical aspect of it. It also gives an interface to the admin personas to deploy rules of routing workloads, controlled cloud bursting, maintaining isolation between workloads without wastage of resources and other checks and balances in place to keep the cost under control.
• Analytics anywhere - Provides a sql analytics engine for on-prem use as well as for use on cloud
• You need to run ad-hoc and Schedulable queries
• You need to minimize cost of analytical queries on clouds - Since its pricing is not associated with the amount of data which is scanned in every execution so the cost does not climb up suddenly on high usages
• Controlled cloud usage - Automatic “cloud burst” of workloads on high demand. Shrinks back to using on-prem cluster
• Sharing of resources (clusters) across multiple teams is transparent to the users
• Isolation of workloads, despite sharing, is carefully designed in the system
• Custom ways in which you want to share the computational resources across teams
• Separate the concerns of the ‘users’ of the resources and the ‘system admins’ of the resources
a) Single pane of glass for multiple clusters, behind the Falarica Query Platform gateway spares the users from bothering about actual connection urls and connectivity to multiple clusters
b) Admin operations like, maintenance, addition of clusters, removal of clusters, addition of catalogs etc can be carried out without affecting the end users
• To control from a central management point the use of on-prem clusters as well as cloud resources and usages
Some of the important problems Falarica Query Platform solves are
1. Provides an analytical engine for on-prem - Standard ANSI sql support for analytics
2. Maximizes or optimizes on-prem resource usages before shifting workloads to the cloud - Overall Cost Efficiency
3. Obviates old style “reservation system” for claiming resources. This is not only insufficient but also leads to wastage of resources - Automatic, intelligent sharing of resources, with sophisticated workload isolation
4. Enables admins to author rules for very custom rules for resource usages
We would be very happy to get in touch with you to understand your analytical workloads requirement and do a poc for you. Please click here to contact us and register for a poc
The main technology that powers Falarica Query Platform are Presto and Kubernetes
For customers whose operations are limited to a single cloud or a single geographical zone, Falarica Query Platform provides multiple values over and above a Presto cluster. Falarica Query Platform is integrated with an internal authorizer and an internal hive metastore. Falarica Query Platform provides auto scale up and down capabilities to save costs. Falarica Query Platform has an integrated UI query interface to execute queries. The UI interface is tied with authentication and authorization too.
An enterprise having a good in-house infrastructure of machines may in all likelihood, would want to fully use the in-house resources before starting to use the cloud resources. Similarly it will be very useful to them if the workload can shift back to using on-premise resources when the resources become available again. A mechanism to transparently do this can be very effective in saving cost of the cloud usages
Enterprises may be working at the same time with multiple clouds. There may be a range of reasons why it would be desirable to use one or more clouds for handling analytic workload needs, along with or without on-prem resources. Some of the reasons could be point in time relative costs, different data locations etc. Please note working with multiple clouds is a norm these days, the question is do you have a single pane of glass through which you can access your clusters in multiple clouds. With Falarica Query Platform the answer is not only “yes” but it is completely transparent to the end user as to how the workloads across the organisation are getting orchestrated on multiple clouds and even multiple regions and zones in the same cloud.
Falarica Query Platform cloudbursts the workload only when the on-prem resources are exhausted or more specifically when on-prem resources are about to reach a certain threshold. The default strategy looks to be very logical if you want to minimize the cloud cost or rather go to the cloud only when absolutely necessary. Custom rules for cloudbursts can also be setup in Falarica Query Platform
Falarica Query Platform keeps a very rich set of statistics about various aspects of the systems and its usages. These information are kept in internal database tables and standard sql can be fired to look at the statistics. Rich reports on query wise usages, helping isolate expensive queries from the mix, team and user wise usages of cloud consumption and on-prem consumption can be easily obtained. Rich insights gained from them can help debug performance bottlenecks in the queries, team wise resource consumption and distribution so that better decision can be taken to tune Falarica Query Platform and place rules which further helps in optimizing the cluster usages and lower the costs
Provides both Authentication and Authorization for queried and admin operations on the gateway. For more details please contact us here.
Falarica Query Platform is an ANSI SQL compliant query engine. This standard compliance allows Falarica Query Platform users to integrate their favorite data tools, including BI and ETL tools with any underlying data source.
Well, if you just take the query running capabilities of all the three systems, then they are at par. Apart from other differences specially with GBQ, like GBQ is a full fledged cloud Data Warehouse product, the main difference is between how the cost is calculated by the three engines for executing queries. In Athena and GBQ the cost is directly proportional to the amount of data scanned i.e. processed while executing the query. With time the number of queries as well as the data size keeps on increasing leading to ever increasing costs. In Falarica Query Platform however, the cost is mainly related to the computational resources which you are using, the actual machine costs. With effective scale-up and scale-in strategies, further optimal usages are guaranteed while keeping the cost under control.
If we are talking about the Presto proxy that comes bundled with the Presto then that proxy has got a very specific use. It acts as a bridge between the real Presto cluster which is behind a firewall and the outside world. So securing the actual cluster behind the proxy seems to be the only paramount reasons so it cannot be compared with Falarica Query Platform as such.
Other kinds of proxies, like load balancers can be used which along with keeping the clusters secured can also do some simple load distribution like round-robin distribution of workload. In real scenarios the desired behavior of the orchestrator may need to be much more customizable owing to needs of different SLAs, teams, data locality and hence preferred clusters, restricting of usage, optimized usage of all the clusters and on and on.
Falarica Query Platform by default uses the the Prestosql version 334 at this point. However, It does not bother the exact version and users have the flexibility to replace their own version of Presto through the Falarica Query Platform’s configurable options. It essentially uses Falarica’s Presto Operator to launch Presto on any Kubernetes clusters. The Presto image can be mentioned in the CRD (Custom Resource Definition) file.
Falarica Query Platform gives an option to specify the Presto docker image and registry. So yes the user can use the Presto distribution of their choice
Falarica Query Platform's gateway component is built on PrestoSQL. However it works with both PrestoSQL as well as PrestoDB clusters.
Falarica Query Platform supports both PrestoSQL and PrestoDB. User can specify image of their favorite Presto distribution.