FabricPool
What Is NetApp FabricPool?
FabricPool is a NetApp ONTAP storage technology that automatically tiers inactive data blocks from an all-flash performance tier to lower-cost object storage, either on premises or in the public cloud. ONTAP presents the flash tier and the object tier as a single logical pool and decides which blocks live where.
FabricPool is a form of storage-based tiering, also called block-level tiering. The storage array moves pieces of files, not whole files, and it keeps control over how that data is read back. That design choice determines how much a FabricPool deployment saves, what happens when you refresh storage, and whether the tiered data is usable for anything outside the array.
Read more about storage pools.
Enterprise data storage vendors have had a complicated relationship with data tiering. While some storage vendors continue to make claims about needing no tiers, most data storage vendors now acknowledge that 70%+ of unstructured file data is cold data and provide options for tiering to lower-cost storage. While lower cost drives were initially the target for the cold tier, many vendors now also offer tiering through their file system to object storage on-premises or in the cloud.
Storage-based data tiering and pools-based techniques to move data within a single architecture or operating system can deliver cost savings for system data such as snapshots. However, storage tiering introduces many limitations when it comes to flexible data policies, storage cost savings, ransomware defense, broader data access and unstructured data management needs.
Read the blog post: What you need to know before jumping into the cloud tiering pool
Download the white paper: Cloud Tiering: Storage-Based vs Gateways vs File-Based: Which is Better and Why?
Learn more about the Komprise path to the cloud for file and object data.
How FabricPool Tiering Works
FabricPool attaches an object store, called the cloud tier, to an all-flash aggregate, called the local tier. A tiering policy set on each volume controls what moves:
- Snapshot-only: tiers blocks referenced only by Snapshot copies once they go cold.
- Auto: tiers any cold block, including blocks in the active file system, after a cooling period of inactivity.
- All: sends data directly to the cloud tier, typically for secondary or data protection volumes.
- None: keeps all data on the flash tier.
ONTAP groups cold blocks into larger objects and writes them to the bucket in its own format. The files are not reassembled in the object store. Only ONTAP can read the tiered data, so applications, cloud services, and other storage systems see a bucket of ONTAP objects, not a set of usable files.
Why Tiering Matters for Unstructured Data
Tiering exists because most file data stops being used soon after it is created. More than 70% of unstructured file data is cold, yet it sits on the same flash as active data. Unstructured data also grows 20% to 150% per year depending on the industry, so the cold share of a flash system keeps getting larger.
Files are harder to tier well than structured records. There is no schema to tell you what a file is, naming is inconsistent, ownership changes, and a single share holds billions of files across departments, projects, and applications. A block-temperature metric tells the array which blocks are idle. It does not tell you whose data it is, what type it is, or whether it belongs in an archive, a retention bucket, or an AI pipeline.
Why Intelligent Tiering Matters More Now Than Ever
Flash is no longer getting cheaper. Gartner projects a 130% increase in combined DRAM and SSD prices by the end of 2026 compared with 2025. TrendForce expected enterprise SSD contract prices to rise 53% to 58% quarter over quarter in the first quarter of 2026, a record quarterly increase. Storage was already a large line item before the price surge. Data storage, backup, and disaster recovery consume more than 30% of the IT budget at 55% of enterprises.
Source: Komprise 2026 State of Unstructured Data Management Report
When flash costs more every quarter, every terabyte of cold data on flash costs more too. Buying more flash is the most expensive way to add capacity. The lower-cost move is to take cold data off flash and keep it accessible. The tiering method you choose decides how much of that saving you keep, and whether it survives your next storage refresh.
Intelligent tiering means tiering driven by analytics about the data itself: age, last access, file type, owner, project, and cost by location. It also means tiering that works across every storage vendor and keeps tiered files usable. Block-level tiering inside one array delivers only part of that.
Disk and object storage prices are rising too. NetApp has raised prices by up to 40% on its hard drive-based storage and by as much as 35% on select Keystone storage-as-a-service tiers. FAS, E-Series and StorageGRID, the platforms most often used as capacity and FabricPool tiers, rose 10% across hardware, software and services. Tiering only pays off fully when you can choose where cold data lands. See NetApp FabricPool for the full comparison.
Source: CRN, November 2026; HostingJournalist, November 18, 2026
Flash prices are not coming back down
Gartner forecasts NAND flash revenue will grow 371.9% in 2026, driven largely by pricing, and expects supply and demand to stay tight through 2027 as AI infrastructure consumes more memory.
Source: Gartner
TrendForce projects NAND flash contract prices will rise another 15% to 20% quarter over quarter in the fourth quarter of 2026, with enterprise SSD the only memory category set to increase faster than it did the quarter before.
Source: TrendForce
TrendForce also expects memory contract prices to remain broadly elevated in 2027.
Source: TrendForce
Waiting for relief is not a plan. Every terabyte of cold data on ONTAP flash is a terabyte you keep paying for at these prices, and every tiered block that gets recalled lands back on that flash.
What Is the Rehydration Penalty?
The rehydration penalty is the cost, time, and capacity consumed when tiered data has to be pulled back to primary storage before an operation completes. With block-level tiering, the tiered blocks mean nothing outside the array, so any operation that needs the whole dataset outside that array forces a recall.
Common data rehydration triggers include:
- Storage refresh or vendor change: moving to a new array or a different vendor requires reading every tiered block back through ONTAP before the data migrates.
- Cloud migration: the objects in the bucket are not files, so moving the data to a cloud file service or another platform requires a full recall first.
- Backup and scanning: backup jobs, antivirus scans, and other tools that read full files pull tiered blocks back, so the backup footprint does not shrink.
Each recall generates cloud egress and retrieval charges, consumes flash capacity you tiered to free up, and adds time to the project. The backup multiplier for primary data is typically 3X to 4X, so a tiering approach that leaves backups unchanged leaves most of the cost in place.
Read: Block-level Tiering vs File-level Tiering
The rehydration penalty is also what makes block-level tiering a form of lock-in. The more data you tier, the more it costs to leave.
Komprise Intelligent Data Management: File-Level Tiering Without Rehydration
Transparent file-based tiering differs from FabricPool in three ways that matter to storage budgets:
- Different policies for different data. Set tiering by share, directory, file type, owner, project or last access, not one volume-level policy based on block inactivity.
- No bringing data back. Tiered files stay in native format on the lower-cost tier. Backups, migrations and refreshes never require rehydration.
- Freedom to switch vendors. Tiered files are readable without ONTAP, so you can change primary NAS, object or cloud vendors without unwinding your tiering.
File-level tiering moves whole files, not blocks, so the tiered data stays usable on its own. Komprise Intelligent Data Management tiers file data from NetApp ONTAP and other NAS and cloud storage using Transparent Move Technology.
- Analyze before you tier. Komprise Analysis indexes metadata across NetApp and other storage into the Global Metadatabase, shows how much data is cold by age, type, owner, and share, and models the savings for each object or cloud destination. The Komprise Flash Stretch assessment applies this analysis to identify how much primary flash capacity you can free.
- Tier whole files transparently. Transparent Move Technology moves cold files to object storage in their native format and leaves a transparent link at the source, so users and applications access files exactly as before. There is no rehydration penalty, and Transparent Move Technology cuts storage costs by 70% or more.
- Keep tiered data usable. Because files land in the object store as files, you access them directly with cloud-native analytics and AI services. Deep Analytics finds the exact files you need, and Smart Data Workflows deliver them to downstream AI and analytics tools without recalling them to primary storage.
- Refresh storage without recall. Tiered files do not depend on the source array, so a storage refresh or vendor change migrates only the active data. Elastic Data Migration moves the rest, with 27X faster NFS migrations and 25X faster SMB migrations using Komprise Hypertransfer.
Here is a summary of file-level tiering benefits over native NetApp tiering, including ransomware protection with object-locked tiered data:
| Evaluation Criteria | FabricPool Block Tiering | Komprise Transparent File-Based Tiering |
|---|---|---|
| What moves | Cold 4 KB blocks inside an ONTAP aggregate | Whole files with metadata and permissions |
| Tiering policy | One of four volume-level policies (none, snapshot-only, auto, all) based on block inactivity | Different policies by share, directory, file type, owner, project, size or last access |
| Policy input | Block temperature inside one system | Deep Analytics queries across every storage system |
| Tier targets | Object stores attached to ONTAP, often StorageGRID or a cloud bucket | Any vendor’s disk-based NAS, object or cloud storage, with a different target per policy |
| Who can read tiered data | Only ONTAP | Any application, user or AI tool, directly from the target |
| User access | Transparent through ONTAP | Transparent. Same path and permissions |
| Bringing data back | Rehydration required for many backups, migrations and refreshes | Never required. Tiered files stay on the lower-cost tier |
| Backup footprint | Tiered data remains in the backup set | Tiered files leave the primary backup footprint |
| Switching vendors | Recall tiered blocks to primary before leaving ONTAP | Replace primary, object or cloud vendors without unwinding tiering |
| Flash price exposure | Hot and recalled data stay on ONTAP flash | Only active data stays on flash |
| Storage coverage | NetApp ONTAP | NetApp, Dell PowerScale, Everpure, Windows and other NAS, object and cloud |
| AI use of tiered data | Not usable outside ONTAP | Indexed and available to Smart Data Workflows, Komprise Intelligent AI Ingest and Universal File MCP |
| Savings modeled before acting | Per-aggregate estimates | Cost modeling across storage and backup with Komprise Flash Stretch |
Independent Evaluation and Measured Value
In the DCIG Pulse for Unstructured Data Management, which evaluated 12 solutions, Komprise ranked first in five of eight vendor criteria. DCIG found Komprise was the only vendor with full support for transparent access and path preservation, and the only vendor with full support for cost savings reporting. DCIG also places data management features built into a single storage platform outside the unstructured data management category.
Source: DCIG Pulse for Unstructured Data Management, November 2026
A Komprise analysis of 10 organizations with 100 TB to 10 PB of NAS data, including NetApp environments, found that Komprise reduces annual storage and backup costs for cold data by an average of 70%. For a 4 PB NAS environment growing 30% a year, that is an average of 57% of overall storage costs, or more than $2.6 million a year.
Source: Quantifying the Business Value of Komprise Intelligent Data Management
AI agents on NetApp data with Komprise Universal File MCP
Komprise Universal File MCP, launched September 29, 2026, gives AI agents and LLMs one Model Context Protocol interface to file and object data on NetApp and every other storage silo, inside the AI tool people already use, such as Claude, ChatGPT, Gemini or any MCP-compatible assistant.
- One interface, one schema. Agents query the Komprise Global Metadatabase instead of loading a separate MCP server and tool set for each storage system, which is the MCP bloat that raises token costs and lowers accuracy.
- Right-sized answers. Progressive response returns metadata first through Transparent File Tables and loads full files only when the task needs them. Deep Analytics filters out irrelevant, outdated, conflicting and duplicate files, and filtering out ROT data improves AI accuracy by over 50%.
- Governed by default. Every request authenticates through existing SSO and returns only data the requesting user is already permitted to see. Komprise audits what data was sent to which AI tool.
- Answers that turn into action. From any result, a user or agent triggers a Smart Data Workflow: confine sensitive files, run a KAPPA data service to enrich metadata, ingest a curated set into an AI pipeline, or export an Apache Iceberg table for Databricks or Snowflake.
- Tiered data included. Files tiered with Transparent Move Technology stay queryable with no rehydration.
Source: SiliconANGLE
Frequently Asked FabricPool Questions
What is FabricPool?
FabricPool is a NetApp ONTAP feature that automatically tiers cold data blocks from an all-flash tier to lower-cost object storage on premises or in the cloud. ONTAP treats both tiers as one pool and moves blocks based on how long they have been inactive and the tiering policy set on each volume. Because it tiers blocks rather than files, the tiered data is readable only through ONTAP.
What is the difference between block-level tiering and file-level tiering?
Block-level tiering, used by FabricPool, moves inactive blocks of data from inside the storage array to object storage and needs the array to reassemble them. File-level tiering moves whole files, keeps them in their native format at the destination, and leaves a transparent link at the source. File-level tiering keeps tiered data usable without the original array, which removes the rehydration penalty during migrations, backups, and cloud access. For more detail, see Block Storage and Storage Tiering.
Does FabricPool have a rehydration penalty?
Yes. Blocks tiered by FabricPool are stored in an ONTAP format, so any operation that needs the full dataset outside the array requires ONTAP to recall them first. Storage refreshes, vendor changes, cloud migrations, and tools that read whole files trigger that recall. Each recall adds cloud egress and retrieval charges, consumes flash capacity, and extends project timelines.
Can I switch storage vendors after tiering with FabricPool?
Not without rehydration. FabricPool-tiered blocks are readable only through ONTAP, so leaving NetApp means recalling tiered data to primary storage first. Data tiered with Komprise stays in native file format on the target, so you can replace primary NAS or change object and cloud vendors without bringing data back.
Why is intelligent tiering more important as flash and SSD prices rise?
Gartner projects a 130% increase in combined DRAM and SSD prices by the end of 2026, so every terabyte of cold data kept on flash costs more each quarter. More than 70% of unstructured file data is cold, which makes tiering the fastest way to free flash capacity without buying more. Intelligent tiering uses analytics on file age, type, owner, and usage to decide what moves, works across storage vendors, and keeps tiered data accessible, so the savings hold through backups, refreshes, and migrations.
Can I use FabricPool tiered data for AI or cloud analytics?
Not directly. FabricPool writes blocks to the object store, not files, so cloud AI services, data lakes, and analytics platforms cannot read the tiered data from the bucket. The data has to be recalled through ONTAP and copied out first. File-level tiering stores whole files in native format, so you point cloud services at the tiered files directly and use them for AI, search, and analytics.
How does Komprise tier NetApp data without a rehydration penalty?
Komprise Intelligent Data Management uses Transparent Move Technology to tier whole files from NetApp ONTAP and other NAS systems to any object or cloud storage in native format. A transparent link at the source keeps access unchanged for users and applications, and the tiered files remain readable at the destination without the source array. Customers cut storage costs by 70% or more, and backup and refresh projects no longer require recalling tiered data. Komprise first shows how much NetApp data is cold and models the savings before anything moves. Learn more about a Storage Refresh Assessment.
Why do I need Komprise if I already use FabricPool?
FabricPool reduces flash used by cold blocks inside one ONTAP system. Komprise reduces the total cost of cold files across every system: whole files leave flash and the backup set, stay readable at the destination, and need no recall during refreshes or migrations. Komprise also gives you one Global Metadatabase across NetApp and non-NetApp storage, which FabricPool does not provide.
How do I migrate off NetApp if I used FabricPool?
Plan for the recall. Blocks tiered by FabricPool must return through ONTAP before data moves to another vendor or the cloud, which needs primary capacity and generates egress charges. Komprise Elastic Data Migration then moves active NFS data 27X faster, and Komprise Hypertransfer moves SMB data 25X faster. Tier cold data at the file level going forward so the next migration skips the recall.
Can AI agents use data tiered by FabricPool?
Not from the object store. FabricPool writes ONTAP blocks, not files, so an agent or cloud AI service cannot read them in the bucket. Files tiered by Komprise stay queryable through Komprise Universal File MCP with no rehydration, and agents receive governed, permission-scoped results with metadata first and full files only when needed.
Can Komprise deliver tiered NetApp data to AI?
Yes. Files tiered by Komprise land in object or cloud storage in native format, so cloud AI and analytics services read them directly. Deep Analytics finds the exact files needed across NetApp and other storage, Smart Data Workflows classify and tag them, and Komprise Intelligent AI Ingest delivers curated data sets with 2X faster data flows.