NextGen lakehouse compute engine
The next-gen analytics engine for
“heavy” workloads.
Trusted by NASDAQ listed enterprises and private Unicorns, backed by Accel. Read MoreFaster10-100x
Lower TCO60-80%
Ready to Go Value360° Interoperability Prod-ready in 10 days
10-100x
60-80%
360° Interoperability Prod-ready in 10 days
Second largest category of IT spend at
over $100B
Fastest growing at 30% YoY
CONTEXT
Data leaders are feeling
the pain.
Lack of compelling compute engine alternatives
Limited competition due to deep barriers to entry: Specialized know-how, Massive capital needs , and Long time-to-market.
Existing platforms are indistinguishable on price - performance reducing the incentive to switch.
Worries around compute ecosystem lock-in
Vendors tie core performance, cost to specific table formats (e.g. Delta or Iceberg), catalogs / governance layers (Unity).
Migrating from one engine’s SQL dialect to another engine’s SQL involves months of effort.
NOT ALL DATA INTELLIGENCE WORKLOADS ARE CREATED EQUAL
Some workloads are
“just different”.
Mission critical | compute intensive | non-discretionary
Purpose-built for the 10% heavy workloads
that drive 80% of Cost , Engineering effort,
and Stakeholder complaints.
Introducing e6data
Amplify ROI, unlock new capabilities on existing data platforms.
SPEED / LATENCY
TOTAL COST OF OWNERSHIP (TCO)
GUARANTEED LATENCY UNDER LOAD
Introducing e6data
Negate any compute
ecosystem lock-in.
Truly format-neutral compute, interoperable with all major open standards.
EXISTING BI TOOL / NOTEBOOK / API / OTHER
(JDBC / ODBC / SQL ALCHEMY)
EXISTING TABLE FORMAT
(HIVE, DELTA, ICEBERG, HUDI)
EXISTING FILE FORMAT
(PARQUET, ORC, AVRO)
EXISTING STORAGE LAYER
(S3, GCS, ADLS, HDFS)
EXISTING CLOUD PROVIDER
(AWS, GCP, AZURE, ON-PREM)
how we built e6data
What we did not do (too) differently.
Columnar processing
Pipelined execution
Cache-friendly execution, Vectorisation
Highly parallel, Distributed execution
Optimal Planning, Optimisation
Data caching
how we built e6data
What we did (very) differently.
Novel fully disaggregated Architecture
No centralised coordinator or driver Lightweight, single-purpose services Granular, independent scaling of services No Single Point of Failure (SPOF)
Decentralised task scheduling, execution
No centralised task scheduling, and coordination Distributed executors pull tasks when free Mitigates challenges of variable task time
Own implementations of the SOTA
But on our unique architecture and distributed processing foundations.