Tech, ExplainedData platforms and AI
Product

Cloud storage cost optimisation

Stop paying for empty cloud disks.

BEFORE ITAFTER ITCloud storagecost optimisationLucidityCOVERSAuto-growAuto-shrinkNo downtimeDisk monitoringreportsthe savingsCloud management platform↗ CLOUD HOSTINGAn agent on each cloudserver measures howfull each disk is,every few minutes.CONCEPTS IT USESObject storageBEFORE ITAFTER ITCloud storagecost optimisationLucidityCOVERSAuto-growAuto-shrinkNo downtimeDiskmonitoringreportsthe savingsCloud managementplatform↗ CLOUD HOSTINGAn agent oneach cloudserver measureshow full eachdisk is, everyfew minutes.CONCEPTS IT USESObject storageBEFORE ITAn agent on each cloud server measures howfull each disk is, every few minutes.↓Cloud storage cost optimisationLucidityAuto-growAuto-shrinkNo downtimeDisk monitoringAFTER IT↓ reports the savingsCloud management platform↗ CLOUD HOSTINGCONCEPTS IT USESObject storage
An agent on each cloud server measures how full every disk is, the software grows or shrinks the disks while the servers run, and the savings go to the cloud management platform.
opens a page inside this one a page in the same scene a page in another scene a concept page a label only, with no page

Cloud storage cost optimisation is software that shrinks and grows the disks a company rents in the cloud as the data on them changes, so that the company stops paying for empty space. The disks in question are block storage volumes. A block storage volume is a virtual hard disk attached to one cloud server, such as the disk that holds a database. The page on cloud computing explains how such servers and disks are rented.

Why is there empty space? A cloud provider charges for the size of each disk, not for the data on it. A 1 TB disk costs the same whether it holds 900 GB or 100 GB. Engineers fear a full disk, because a full disk can crash the application on it. So they usually create disks much larger than they need. They also rarely make a disk smaller, because the big clouds do not let a disk be shrunk. To shrink one, the engineer must create a new, smaller disk, copy everything across, and stop the application while doing it. As a result, much of the disk space a company rents is often empty.

How does the software work? A small agent on each cloud server measures how full each disk is, every few minutes. When a disk passes about 75% full, the software adds space while the server keeps running. When data is deleted and the disk is mostly empty, the software gives space back. To make shrinking possible, it builds each disk out of several smaller disks, which the server's operating system sees as one. It can then add or remove the smaller disks one at a time.

Take, for example, an online education company in Gurugram that runs on 300 cloud servers. Their disks add up to 400 TB, but they hold only 140 TB of data. The company pays about ₹7 for each GB of disk every month, which comes to about ₹28 lakh a month. After the software takes over, the disks are kept about three-quarters full, so they add up to about 190 TB. The monthly bill for disks falls to about ₹13 lakh. In February, students upload far more video answers before their board exams. The software grows the affected disks overnight, and nobody has to be woken up.

This product does one narrow job. A cloud management platform looks after the whole cloud account, including the full bill, the servers and the settings. Moving old files to cheaper object storage is another way to save money. But object storage is not a disk that a server can run a database on. The software adjusts only the disks, so the servers and virtual machines that use them keep working as before. The broader question of where data should be kept is covered in data storage and backup.

Tech, Explained is a part of ingrid.fyi. Every page explains one piece of technology in plain words.