India’s Leading Domain & Web Hosting Provider
Support
Domain & Login
Blog
Cloud Storage Optimization: Tier Objects, Retune Volumes, Drop Copies
Storage optimization reduces the cost of cloud object, block and file storage by matching each dataset's storage class, performance tier and retention to how it is actually read. NTSPL analyses access logs and volume metrics on AWS, Azure and Google Cloud, then applies lifecycle tiering, volume-type retuning and duplicate removal while modelling retrieval fees and minimum storage durations.
Delivered on
In scope
Why Hot-Tier Data and Overprovisioned Volumes Dominate Storage Bills
Storage is the cost line that only grows. Compute gets scaled down at night; data is written once and kept forever, in the class it was created in. Typical Indian estates we review hold 60-80% of object data in the hot tier although under 10% of it is read in any 30-day period, keep EBS gp2 or Premium SSD volumes provisioned for IOPS they never reach, and store the same media or log files three times across buckets, regions and backup copies. The cost per GB is small, so the problem stays invisible until storage passes 25-30% of the bill.
NTSPL works from evidence of reads, not file ages. We enable S3 Storage Lens, Azure Blob last-access tracking or Cloud Storage access logs, profile prefixes and containers over 30-60 days, and model every proposed transition against minimum-duration charges, per-object fees and retrieval costs before writing a single lifecycle rule. Block volumes are retuned on measured p99 IOPS and throughput, never on the size someone picked at launch.
When Storage Has Become a Cost Problem Worth Solving
Signals you need Storage Optimization
- Storage exceeds 25% of the monthly cloud bill, or grows faster than 3% a month.
- More than half of object storage sits in S3 Standard, Blob Hot or GCS Standard with no lifecycle rules.
- EBS gp2, io1 or Azure Premium SSD P-series volumes make up most block capacity.
- Media, document or log data is copied into multiple buckets or regions without a stated purpose for each copy.
- Application teams cannot say how long data must be kept, so nothing is ever deleted.
When to choose something else
- The expensive storage is backup vaults and snapshot chains: that is Backup Cost Optimization, which handles retention and vault tiers.
- Cost sits in data warehouse or lakehouse storage driven by query patterns: Analytics FinOps tunes partitioning and table formats.
- Total storage is under 50 TB: enable native intelligent tiering and gp3 defaults yourself; the analysis will cost more than it saves.
Object, Block and File Layers in Scope
Object storage tiering
S3, Azure Blob and Cloud Storage prefixes and containers profiled by read frequency, object size and age, with lifecycle or automatic tiering applied.
Block volume retuning
gp2 to gp3, io1/io2 provisioned IOPS trimmed to p99, Azure Premium SSD to Premium SSD v2 or Standard SSD, Persistent Disk to Hyperdisk or balanced PD.
File shares
EFS, FSx, Azure Files and Filestore capacity and throughput modes reviewed; infrequent-access classes enabled where read patterns support them.
Duplicate and derived data
Repeated copies across buckets and regions, intermediate processing outputs and stale exports identified and either deleted or consolidated.
Retention definitions
Retention periods captured per data class with business and legal owners, then encoded as lifecycle expiry rather than left open-ended.
Incomplete and versioned objects
Incomplete multipart uploads, noncurrent versions and soft-deleted blobs cleaned up through lifecycle rules.
How Storage Optimization Works, Step by Step
Access-Evidence Storage Tiering Model
Access telemetry and inventory reports build a read-frequency profile per prefix or container, which a cost model compares against every eligible storage class including minimum-duration and retrieval charges. Approved transitions are encoded as lifecycle rules and volume modifications, then verified against billing.
Select any component
Access-Evidence Storage Tiering Model: click any of its 9 components to see its role and the design choice NTSPL makes for it, or trace the 7-step primary path.
Storage Classes and Tiering Tools by Provider
gp2-to-gp3 conversion is usually the fastest win on AWS: it runs online, and gp3 decouples IOPS and throughput from size so large volumes stop paying for performance they do not use.
Cold tier sits between Cool and Archive with a 90-day minimum and online access, which fits compliance data that is read a few times a year but must open instantly.
Autoclass moves objects between classes based on access without retrieval or early-deletion charges, so NTSPL prefers it for buckets with unpredictable access and uses lifecycle rules where patterns are known.
Tiering Rules That Avoid Retrieval Surprises
Automatic tiering or explicit lifecycle
Use S3 Intelligent-Tiering or GCS Autoclass where access is unpredictable and objects are mostly larger than 128 KB. Use explicit lifecycle rules where patterns are known, such as logs read for 30 days then never, because rules avoid per-object monitoring fees. On Azure, combine last-access tracking with lifecycle rules. NTSPL decides per prefix, and large estates usually end up with both approaches.
Minimum-duration traps
Every cold class charges a minimum storage duration: S3 Glacier Deep Archive 180 days, Azure Archive 180 days, GCS Archive 365 days. Data deleted or rewritten earlier is billed as if kept. Never push data into a cold class if its retention or rewrite cycle is shorter than the minimum. NTSPL checks the overwrite and delete rate of every prefix before choosing a class.
Volume type selection
Size gp3, Premium SSD v2 and Hyperdisk performance to observed p99 plus 30%, not to the old volume's provisioned figure. Keep io2 or Ultra Disk only for databases whose p99 latency target is under one millisecond. Large gp2 volumes over 1 TB usually save 20% or more on conversion alone; small boot volumes save little and can be left for routine rebuilds.
Copies and regions
Every copy should have a purpose: serving, DR or compliance. A DR copy in AWS Hyderabad, Azure South India or Google Cloud Delhi can sit in an infrequent-access class because it is read only during failover. Analytics extracts duplicated into separate buckets usually should not exist. NTSPL lists every copy with its purpose and owner, and deletes the ones nobody claims.
Prefix-by-Prefix Storage Optimisation Plan
Telemetry enablement
Storage Lens advanced metrics, Blob last-access tracking or GCS usage logs enabled; volume metrics exported.
Access profiling
30-day minimum observation per prefix, container and volume; retention owners interviewed for top 30 data classes.
Cost model and rule design
Class and volume targets modelled including retrieval and minimum-duration costs; rules drafted as code.
Volume retuning
Online volume type and performance changes applied in application batches with latency checks.
Lifecycle rollout and verification
Rules applied prefix by prefix, largest first, with retrieval-fee watch for 30 days after each.
What you receive
Storage Efficiency Indicators
| KPI | Typical target |
|---|---|
| Storage cost reduction on in-scope data | 30-55% |
| Object data in hot tier with no reads in 30 days | Below 10% |
| Retrieval and early-deletion fees | Under 5% of storage saving |
| Block capacity on current-generation volume types | Above 90% |
| Data classes with defined retention | 100% of top 30 by cost |
Tiering Mistakes That Raise the Bill
| Risk | How NTSPL prevents it |
|---|---|
| A batch reprocessing job reads an archived prefix and triggers large retrieval charges. | Data owners confirm reprocessing patterns, and archived prefixes used by any job are kept in instant-retrieval classes. |
| Lifecycle expiry deletes data that a regulator or contract requires. | Expiry rules are only written for data classes whose retention owner has signed off; legal-hold containers are excluded by tag. |
| Per-object fees outweigh savings on buckets full of small files. | The small-object check excludes prefixes where the median object is under 128 KB or recommends compaction first. |
S3 Intelligent-Tiering versus lifecycle rules: which should we use?
Use Intelligent-Tiering when you cannot predict access: user uploads, shared data lakes, media libraries where old items occasionally trend again. It moves objects automatically with no retrieval fees, for a small per-object monitoring charge. Use lifecycle rules when access is predictable: application logs, daily exports, transaction archives that are read heavily for weeks and rarely after. Rules have no monitoring charge and can move data to Deep Archive or delete it. Most estates need both: NTSPL typically puts 30-40% of capacity under Intelligent-Tiering and the remainder under explicit rules.
Key facts: Storage Optimization
Storage Optimization in Practice
2 Storage Optimization scenarios, from News app (multilingual mobile news and short video) and Health-tech (teleconsultation and e-pharmacy platform), plus related platform case studies.
Regional news app halves Cloud Storage cost on 1.9 PB of reader video and logs
Teleconsultation health-tech moves 380 TB of gp2 to gp3 and tiers consult recordings
Regional-language OTT service halves video start time and cuts origin egress by 82%
News channel moves 1.4 PB of footage from LTO tape to tiered Blob Storage
Storage Optimization: Frequently Asked Questions
What does a storage optimisation engagement cost and how long does it take?
Ten to twelve weeks for an estate of 200 TB to 5 PB, driven mainly by the 30-day observation needed before tiering. NTSPL prices it as a fixed fee or as a share of verified savings. Volume retuning savings usually appear within the first month; tiering savings build over the following two to three months.
Does converting gp2 to gp3 cause downtime?
No. EBS Elastic Volumes changes type, IOPS and throughput while the volume stays attached and in use. Performance can be briefly variable during optimisation, so NTSPL runs changes in batches outside peak hours and checks application latency after each batch.
How does the DPDP Act 2023 affect storage optimisation?
The Act expects personal data to be erased once its purpose is served, unless another law requires retention. That makes retention definitions a compliance task, not only a cost task. NTSPL records the purpose and legal retention basis for each data class, so lifecycle expiry rules support both erasure obligations and lower storage cost.
Will archived data still be available when we need it?
Yes, with different access times. Glacier Instant Retrieval, Azure Cold and GCS Coldline open in milliseconds. Glacier Flexible Retrieval, Deep Archive and Azure Archive need a restore that takes minutes to hours. NTSPL puts anything an application might read without warning into an instant-access class.
Should we keep a second copy in another Indian region?
Only where DR or a regulator requires it. If you do, use an infrequent-access class for the copy, since it is read only during failover. Pairs such as AWS Mumbai and Hyderabad, Azure Central India and South India, or Google Cloud Mumbai and Delhi keep both copies within India.
What about versioned buckets and soft delete?
Versioning and soft delete protect against mistakes but can silently double storage. NTSPL adds lifecycle rules that expire noncurrent versions after a defined period, typically 30-90 days, and cleans up incomplete multipart uploads, which are invisible in most console views but still billed.