Delete unused EFS file systems

EFS is a shared network file system — think of it as a network drive that many servers can use at once. The thing finance needs to know is that it's billed on the most expensive common storage meter AWS offers: roughly $0.30 per gigabyte per month, compared with about $0.08 for an EBS disk and around $0.023 for S3. You pay for everything stored on it, every month, regardless of whether a single byte is ever read again. There is no "off" switch — the only way the cost stops is when someone deletes the data or the whole file system.

This finding flags file systems that have had effectively no activity for several weeks — no servers connected, no reads, no writes. In practice these are leftovers: a shared workspace from a project that wrapped, a staging area from a migration that finished, a file share for a team that has since moved to something else. Each is a fixed monthly charge on premium-priced storage with nothing using it. A 2 TB leftover is around $600 a month — over $7,000 a year — for data no process has opened in months.

From a budgeting standpoint EFS is awkward because it grows silently. No one approves a fixed cost the way they would for a reserved server; the number just creeps up under "EFS" as old files accumulate and old file systems are never closed out. The right framing is that an idle, full-priced file system is both a direct cost to reclaim and an ownership signal — it means storage exists that no team is actively accountable for, and it should either be retired, moved to a cheaper tier, or formally owned.

Sam is the finance partner embedded with a platform team. At the monthly cost review she notices the EFS line has been drifting up and asks a question that's now standard on the agenda: "How much of our EFS spend is on file systems with no activity in the last month, and who owns each one?" The dashboard answer is three idle file systems totalling about $410 a month — modest in absolute terms, but it's premium-priced storage doing nothing, and it's been there for several months running, so the cumulative is into four figures.

The conversation isn't technical. Sam doesn't ask about mount targets or IOPS. She asks three things: which project each one belonged to, whether that project is still live, and whether anyone has touched the data in the last 90 days. The engineering lead confirms two are leftovers from closed projects and one is a genuine quarterly-archive share that only wakes up at quarter-end. That's enough to act — retire the two dead ones, and move the archive share to a cheaper tier rather than deleting it.

Two months later the EFS idle line shows around $60 — a single legitimately-infrequent share, now on a low-cost tier with sign-off. Sam knows that's the right floor, and she knows that if the number climbs back above a couple of hundred dollars, it means file systems are being created for projects that then close without anyone retiring the storage. The dollar amount is a leading indicator of a storage-ownership gap; the cleanup itself is almost incidental.

The per-resource impact is larger than most storage findings because of the price tier. EFS Standard is the priciest common storage AWS sells, so a single forgotten file system can quietly be a four-figure annual charge on its own. In aggregate, idle and mis-tiered EFS often lands at 1-3% of cloud spend in environments that lean on shared file systems — not enough to move the quarter alone, but enough to keep landing in "unexplained variance" when budgets are tight.

The biggest reclaimable number is usually not the dead file systems — it's the live ones with no lifecycle policy. Data that's genuinely retained but rarely read is sitting on a meter ~37 times more expensive than it needs to be. Moving it to IA or Archive cuts that cost by 90-97% without deleting a byte or asking anyone's permission to lose data. That makes this category unusually attractive: a large chunk of the saving carries essentially zero business risk.

The third impact is on chargeback credibility. EFS grows silently with no provisioning event, so an untagged, ownerless file system is exactly the kind of line that breaks the chargeback model — real cost that can't be assigned to a business unit. Every untagged gigabyte here is a dollar absorbed centrally or argued over. Retiring or right-tiering with a named owner is less about the saving and more about keeping the chargeback report defensible.

Finally, it's a leading indicator. EFS spend that grows month over month with flat activity almost always means storage is outliving the projects that created it — no closeout step, no lifecycle policy. That same gap tends to predict other storage waste (orphaned snapshots, unattached volumes, abandoned S3 buckets) trending up too. Watch the idle-and-mis-tiered EFS number as a signal of storage discipline, not just a target to zero out.

EFS is shared cloud storage priced at the top of the range — several times the cost of the disk attached to a server, and more than ten times the cost of object storage. When a file system is left behind by a finished project, it keeps billing on every gigabyte it holds, indefinitely, even though nothing reads or writes it.

This is a hygiene and accountability issue wearing a cost label. Each flagged file system is premium-priced storage with no owner and no active value. Reclaiming it is low-risk and immediately saves money, but the more durable win is the norm it sets: storage gets retired or right-tiered when its project ends, rather than accumulating quietly on the most expensive meter the cloud provider sells.

At one company the quarterly cloud review used to carry a quiet line: a few thousand dollars a year of EFS storage on file systems with no activity. Small relative to the total, but it never went down, and occasionally a new one appeared after a project wrapped. The exec sponsor stopped asking about the dollar amount and started asking about the shape: "Why does storage outlive the project that created it? Who is supposed to close it out?"

Within two quarters the answer changed. Project closeout now includes a storage decision — retire it, or move it to a cheap tier with a named owner — and the idle-EFS line dropped to a small residual of intentionally-archived shares. The cleanup stopped being a recurring agenda item. The exec hadn't asked the team to chase gigabytes; she'd asked them to fix the missing closeout step the gigabytes were exposing.

That's the right outcome state. "Zero EFS" is the wrong goal — file systems are useful. "No premium storage outlives its purpose without an owner" is the right one. The cost report stops being an action item and becomes a confidence signal.

The dollar amount can be larger than people expect because EFS is premium-priced, but it's still tracked mainly for what its size and trend say about discipline. A small, flat number means storage gets retired or right-tiered when projects end. A growing number means the inverse: file systems outlive their purpose, no one enables the cheap tiers, and the same closeout gap almost certainly applies to other storage categories that won't surface as cleanly on the dashboard.

There's a second-order risk: forgotten file systems with stale ownership often hold real data — sometimes regulated data — on storage attached to no application and owned by no one. That shows up as an audit and security finding as readily as a cost finding. So this category sits at the intersection of cost, data governance, and operational hygiene. The CFO cares about the first, the CIO about the second, and both should care about the closeout discipline underneath.