Skip to main content
StickyVeeam Oxford Style Debate -Episode 2

Oxford Style Debate – Episode 2: Tape Is Dead. Or Is It? Let’s Debate.

  • June 23, 2026
  • 53 comments
  • 529 views
Show first post

53 comments

matheusgiovanini
Forum|alt.badge.img+9

What can I say, tape is an awesome theoretical solution that is air-gapped, different media, off site storage, etc.

but all these wonderful features fail when you need a quick restore, or a historical search, or need to migrate all your tapes due to technology change.

i was a fan, but since the day I suffer the tape world, I decided to migrate to HDD, S3, cloud, etc.

we keep having tapes, but they are the last resort, and I don’t really trust them  anymore.

cheers.

I totally understand your point. I felt some of this pain recently during a large recovery from tape.

In my case, we had around 130 VMs in the same media pool, so the restore was not simple at all. Loading/unloading tapes, finding the right restore points and planning the restore order made everything slower than a disk restore.

But we followed a priority strategy: AD, databases, logistics systems and critical services first. The goal was not to restore everything at once, but to make the environment operational again. Medium and low impact VMs came after.

And this is the point for me: it was not perfect, but it worked.

The backup server was compromised, repositories were compromised, and storage replication was compromised. Without the offline tapes, we would basically have nothing to restore from.

So yes, tape is painful sometimes. It is not perfect and it is definitely not the fastest option. But when it is well planned, with good media pools, media sets, rotation, maintenance and proper handling, it can still save the environment.

Another important point is cost. In projects, when we identify that an environment is not really following a 3-2-1 strategy, we usually recommend different ways to improve resilience: immutable repositories, object storage, backup copy, tape, offsite copies, etc.

But not every environment can adopt the ideal strategy right away. Some still rely on a very simple local backup strategy, and cost has a huge weight in these decisions.

So for me, tape is not about convenience. It is about having something to restore when everything else is gone, without making the cost impossible for the customer.

Matheus Giuliano Giovanini | Senior Backup & Data Protection Engineer | Tech Lead | Veeam Legend | Blog: techdirectarchive.com
    Dynamic
    Forum|alt.badge.img+14
    • Dynamic
    • Veeam Vanguard
    • June 29, 2026

    Sending the challenge back over to Germany again with ​@Dynamic 

     

    Thanks for challenging me. I’m in Team Tape isn’t dead! Why?


    Let me tell my thoughts

    • Real air-gap, not a software promise: A cartridge in a vault can't be encrypted, deleted, or accidentally “de-immutabled” (is this the correct word?) via an API. Object-lock is software, and software could have bugs, misconfigurations and sadly sometimes admins/users with too many privileges.
    • Regulatory reality (often a Germany-specific thing): We have many customers who often request a 3-10 year retention. Also in some financial services, BaFin requirements (MaRisk, outsourcing) often explicitly call for a physically separated copy.
    • The "CEO wants the last backup under his pillow" factor: Not a technical argument, but very real: if your chief justice wants to physically hold the last backup, that's stakeholder management and the best architecture fails without money given.

    Where "modern" solutions fall short in practice

    • Object storage stays online: and online means reachable by default. Ransomware groups now target IAM keys and API access directly, not just production VMs.
    • Unseen or Hidden costs: Ingress is cheap, mass-restore from cold tiers can take weeks and get expensive fast.

     

    Conclusion

    • Tape handling costs staff time - manual swaps, labeling, vault logistics (taken offline, transport to a vault).
    • Tape never belongs at the front of the recovery chain for RTO-critical systems — it's simply too slow for that role.
    • That's exactly why the original statement is a false conflict: it's not "tape OR object storage," it's 3-2-1-1-0 -> tape delivers the cheap, air-gapped long-term copy; object/disk delivers the fast restores.

    For sure, the Tape discussion also depends on your customers requirement. I wouldn’t recommend Tapes in every customers solution - probably the best answer would be, it depends 😉

     

    Let’s throw the ball to another continent back, i call out for DJ JP! ​@Jean.peres.bkp 

    Markus Hartmann | Veeam Vanguard | Veeam Legend 2024 | VMCE+ 2026 | VMCA 2024 & VMCE 2024 | VMware Certified Implementation Expert - DCV | https://markushartmann.blog/
    H

    Ha. You did your research well, asking me about tape ​@Madi.Cristil.

    (EDIT) I consider myself AGAINST the statement “tape is dead”.😉

    (Double negation issue… 😉)

    Actually, I tried to convince ​@Anton Gostev in the early days (~2011+) on several events in Germany to integrate tape into VBR. I took me until V7 to see it in the product. 😁

    In the following VeeamONs Anton put all advantages into his presentations and celebrated the integration.

    I’ll summarize the advantages here:

    Security and Ransomware Protection

    • True Air Gap: Once a tape is ejected from the drive, it is physically disconnected from any network or host. No malware, ransomware, or compromised admin account can reach or alter the data. Disk repositories (even immutable ones) remain electrically and logically reachable.
    • WORM / Strict WORM: LTO WORM media and the Veeam Strict WORM mode prevent any append, overwrite, or deletion at the media level. Disk immutability is software-enforced (e.g., XFS reflink, Hardened Repository, Object Lock) and depends on a healthy OS and clock.
    • Offsite portability: Tapes can be transported to a vault (Iron Mountain, bank safe, second site) without bandwidth or replication infrastructure.

    Cost and Capacity

    • Lower cost per TB: LTO-9 (18 TB native, 45 TB compressed) typically costs a fraction per TB compared to enterprise SAS/SATA disks or SSDs, especially when long retention is required.
    • No idle power: Tapes in a slot or vault consume zero watts. Disk arrays draw power and cooling 24/7, even for cold data.
    • Scaling by media, not by chassis: Capacity expansion means adding cartridges, not buying additional shelves, controllers, or rebuilding RAID groups.

    Longevity and Reliability

    • Media lifetime: LTO tapes are specified for 30 years of archival storage under proper conditions. HDDs typically degrade after 5 to 7 years, SSDs suffer from charge loss when powered off for extended periods.
    • Bit error rate: LTO-9 has an uncorrected bit error rate of 1 in 10^20, which is 10,000x better than enterprise SAS HDDs (1 in 10^16) and most SSDs.
    • No silent corruption from firmware/controller: Tape has no complex RAID controller, cache battery, or firmware logic that can corrupt large datasets at once.

    Veeam-Specific Advantages

    • GFS Tape Jobs: Native Grandfather-Father-Son retention for long-term archive without inflating the performance tier.
    • Tape as a 3-2-1-1-0 component: Tape covers the "1 offline copy" requirement of the 3-2-1-1-0 rule cleanly, which disk-based immutable repos cannot fully satisfy.
    • Independent restore path: A tape can be read in any compatible LTO drive worldwide, even without the original Veeam infrastructure (via standalone tools or a fresh VBR install with catalog import).

    Regulatory and Audit

    • ISO 27001, NIS-2, BSI Grundschutz: Auditors generally accept physically separated tape copies as proof of offline backup. Disk immutability often requires additional documentation of configuration, patch level, and access controls.
    • Long retention requirements (e.g., 10+ years for pharma, finance): Tape is the only medium where the TCO remains predictable over such timeframes.

    Operational Aspects

    • No rebuild times: A failed cartridge affects only that cartridge. A failed disk in a large RAID6/RAID60 set can trigger rebuild times of days with elevated risk of secondary failures.
    • Predictable sequential throughput: LTO-9 delivers around 400 MB/s native per drive, which matches or exceeds many SATA HDDs and is ideal for large sequential Veeam backup files.

    Limitations to be aware of (for fairness)

    • Random access and restore granularity are slower than disk. You often have to stage to disk first.
    • Instant Recovery, SureBackup, and Veeam DataLabs require disk, not tape.
    • Tape handling requires either an autoloader/library or manual operations.

     

    Honestly speaking, I always size to in a way that you never have to restore from tape under normal circumstances. Tape is merely meant for the “sum of all fears” event. But then, it has your back.

    In Germany or even DACH (Germany, Austria, Switzerland), tape is often requested in projects and according to the advantages specified above it is a valid option I always present in strategic discussions.

    Even those who dismiss tape as "totally 90s", a relic of a long-gone century, should take note: if you are leveraging AWS Glacier or Azure Archive Tier, you are already using tape backups, hosted at the largest tape operators on this planet. 😎

    I challenge ​@benharmer for an idea from the other side (of the planet, maybe not the topic). 

    Great Post!…..I'm waiting for the application plugin to integrate better with Tape for LTR…...it will surely come in release 13.1🤞

    Terms & ConditionsAccessibility statement