Skip to main content

ICLR Honorable Mention ICLR 2024 Awards Honorable Mention for Granica Research | Read ICLR's blog →

Get a demo

Cloud Cost Management Platform

Shrink cloud costs for really big data

Granica Crunch is the world’s only data lake compression service. Granica’s cloud cost management platform helps data teams significantly improve their cloud data lake storage efficiency to shrink costs by up to 80% every month.

See it in action

The Data Cost Management Problem

It’s hard to keep the costs of data lean while experimenting with AI. 

economics

AI Data is Expensive at Scale

Even with low cost cloud object storage powering your data lakes, a single PB of data has an annual at-rest cost of nearly $300k/year.

Tiering and archival are not effective

Tiering and Archival Aren't Effective

Moving training data into colder, less expensive classes doesn't help much since training data still gets regularly accessed.

Introducing Granica Crunch

Cloud Cost Management for Data, at the Scale of AI

Granica Crunch shrinks data lake at-rest costs by up to 80%, saving millions annually for typical data-intensive companies. It does this by losslessly compressing data lakes with state-of-the-art algorithms while integrating seamlessly with applications and pipelines.

Cloud Cost Management Use Cases

Optimize costs to store data

Control and lower the at-rest storage costs by up to 80% for large-scale data sets used for analytics, ML and AI and stored cloud data lakes.

Optimize costs to move data

Control and lower cross-region transfer costs by up to 80%, to address AI-related compute scarcity, compliance, disaster recovery and other requirements.

Speed training and analytics

Accelerate the performance of any query, application or process bottlenecked by network or IO bandwidth.

Granica Crunch in Action

With Granica Crunch can make data lean and efficient to control costs and improve ROI on analytics and AI.

Learn how Granica Crunch can help you manage AI data costs.

Get a demo

Flatten your cloud cost curve

Granica’s cloud cost management platform  helps reduce your unit economics cost to store data while avoiding the performance, availability, and highly variable read cost penalties associated with tiering and archival. Capture recurring savings relative to your native baseline each and every month, not just once. Ten petabytes of data typically translates into an annual of cash savings of >$1.3M, scaling as your data grows.

  •   Lower your storage unit economics.
  •   Get >$1.3M annual savings per 10 PB (typically).
A graph showing the monthly Amazon S3/CGS cost savings with Granica Crunch

Data types supported

Clickstream, Logs, LiDAR, Image, and more

Granica supports a wide range of AI/ML/analytics data types, with lossy and fully lossless compression variants. We can further customize the platform for your specific use case.

use-case-clickstream

Clickstream/Logs

use-case-tabular

Tabular

use-case-lidar

LiDAR

use-case-satelite

Image

Related Articles

Top Cloud Cost Management Tools: 2024 Review

Top Cloud Cost Management Tools: 2024 Review

Cloud Cost Optimization: The Ultimate Guide

Cloud Cost Optimization: The Ultimate Guide

Google Cloud Cost Optimization Guide

Google Cloud Cost Optimization Guide

Get started in minutes

1
Install and enable Granica Crunch

Deploy into your AWS or GCP cloud environment and ensure the Granica CLI (command line interface) is accessible.

2
Specify buckets to be crunched

Specify buckets that are eligible to be crunched (i.e., compressed) via policy, then run ‘granica crunch ’ to begin crunching data.

3
Work with crunched data

Enable custom applications to read and write through Granica via our SDK and/or SDKless solutions.

bash

g ~/ granica deploy

Success!

g ~/

cyberdust.py

g ~/ granica crunch s3://cyberdust

Crunching started...

g ~/

Screen data at-rest in the background

Work with crunched data

Frequently Asked Questions:

How much cost reduction does Granica Crunch typically deliver?

Granica Crunch typically reduces training data-related storage and access costs by 25-60%, depending on your file types and access patterns. If you’re storing 10 petabytes of training data in your cloud data lake, that translates into more than $1.3 million per year, growing as your data grows, assuming a 50% compression rate.

Does Granica Crunch affect read/write latency?

Granica Crunch is extremely fast, adding only ~0-100 ms of latency (~0 ms for read cache hits), and delivers sustained throughput of up to 1.5 GB/s (not Gbps) per node for transparent access to your training data.

How does Granica Crunch compare to alternatives?

AI data is hot data, and Granica Crunch delivers the savings efficiencies you need without the trade-offs:

  1. Vs. DIY compressions: Granica Crunch removes data pipeline complexity, risk, and latency/throughput challenges.
  2. Vs. archival and tiering to cooler/colder classes: Granica Crunch enables training data to remain instantly accessible and highly available, with no data access charges.

Save. Re-allocate. Accelerate.

Harness the inefficiency in petabyte-scale training data to fund the capture and use of more data to build more accurate models. Granica’s cloud cost management platform optimizes IT spending while transforming data management into a strategic powerhouse, freeing you to focus on tech innovation. 

Read the docs
Get a demo