Lower your Data WarehouseSnowflakeAzure SynapseBusiness IntelligenceGoogle BigQueryAmazon RedshiftDatabricksAmazon AthenaData Lakehouse
Key Features
Everything you need for analytical query processing, transformation pipelines, and data apps.
ANSI-SQL Compatiblity
Full SQL support including complex joins, views, expressions, window functions - you name it !
Low Latency
Directly feed your data intensive apps from AkashX, no need for additional middle tiers to backend the apps.
High-Concurrency
Concurrent analysts, concurrent dashboards or concurrent pipelines, or all at once !
Elastic Scalability
Start small and scale as you grow in terms of data volume, workload volume, or both.
Support for Iceberg, Hudi
We support open file formats - Apache Iceberg, and Hudi - without need for any data migration.
Easy migration
Switching is a breeze, no rewriting pipelines, dashboards, or scripts - just change the endpoints !
What is Storage Acceleration?
AkashX takes the well known concept of pushing compute closer to data to a whole new level!
TPC-H Benchmark
Data size: 100 GB
Number of queries: 22 queries
Instance specifications:
Snowflake - Medium Warehouse (4 credits * $2)
AkashX on Google Cloud - e2-standard-16 ($0.62)
TPC-H Benchmark
Data size: 100 GB
Number of queries: 22 queries
Instance specifications:
Redshift - Tier 8 RPU ($2.88)
AkashX on Google Cloud - e2-standard-16 ($0.62)
TPC-H Benchmark
Data size: 100 GB
Number of queries: 22 queries
Instance specifications:
Snowflake - Medium Warehouse (4 credits * $2)
AkashX on Google Cloud - e2-standard-16 ($0.85)
TPC-H Benchmark
Data size: 100 GB
Number of queries: 22 queries
Instance specifications:
Snowflake - Medium Warehouse (4 credits * $2)
AkashX on Google Cloud - e2-standard-16 ($0.85)
Architectural Features
AkashX is architected for fast, scalable, concurrent, and fresh data analytics.
Vectorized query execution
Single-Instruction-Multiple-Data (SIMD) and Vector Instructions are leveraged for highly efficient query processing.
Columnar data storage
Data is stored in column-major format to speedup analytical queries and compressed loading.
Massively Parallel Processing (MPP)
Analytical queries are broken into many fragments and efficiently delegated to every thread available for high throughput, massively parallel execution.
Unbounded scalability
Shared-nothing architecture allows the storage layer to scale infinitely and topology-aware query query execution allows minimal data shuffling.
Adaptive replication of hot data
Hot and large chunks of data are proactively redistributed to optimally load balance across the cluster.
Join acceleration
Toplogy aware join execution allows pushing down the query fragments to storage servers and minimizing the inter-instance result shipping.