- BFSIBanking, Financial Services & Insurance
- HealthcareComing soon
- ManufacturingComing soon
- Retail & E-CommerceComing soon
- Endpoint Management
- UEM & MDM
- RMM & Patch
- Endpoint Protection
- Remote Access & Support
Endpoint Management
- Security
- Backup & Cyber Resilience
- Backup & Recovery
- SaaS Backup
- Disaster Recovery
- Cyber Recovery
Backup & Cyber Resilience
- Identity & Access
- Privileged Access Management
- IAM, SSO & MFA
- Identity Governance
Identity & Access
- Network Security & SASE
- Firewall & Network Security
- SASE & SSE
- Zero Trust Access
- Secure Web & DNS
Network Security & SASE
Every product here backs things up. What separates them is the day you actually need a restore — at volume, under pressure.
A backup product is four decisions: what it can protect, where it runs, how fast and how granular the restore is, and what unit it bills by. Vendors compete on the first; buyers suffer on the third and fourth.
Commvault publishes a SaaS list of $58.50–90 per TB per month. Veeam is reported at $250–450 per workload per year. Neither number includes the storage the backups sit on — and that is frequently the larger bill.
Already decided — before the demo
Still yours to weigh
What backup and recovery software actually is
A product that copies your workloads — VMs, physical servers, databases, file shares, cloud instances, laptops — on a schedule, keeps the copies for a retention period, and can put them back: one file, one mailbox, a whole VM booted straight from the backup, or a thousand VMs in an afternoon. The copying is table stakes. The restore is the product.
Four things decide the purchase. What it can protect — the workload list varies more than you expect (Kubernetes, NAS at scale, Proxmox, endpoints). Where it runs — software you install, an appliance you rack, or a service with nothing to run. Restore granularity and speed — item-level, full VM, or instant recovery that boots from the repository. The meter — per workload, per TB front-end, per TB after dedupe, per socket, per device. SaaS data (Microsoft 365, Google Workspace, Salesforce) is its own purchase: the SaaS backup guide.
The most common mis-purchase
A backup that succeeds every night and a restore that fails the first time it is needed — because nobody tested it at volume, over the real network, from the real repository. Buy the restore you have rehearsed, not the backup you have scheduled.
Often confused withSaaS Backup — your data in someone else's cloud; they are not backing it up for you →·Disaster Recovery — restoring operations, not only data →·Cyber Recovery — when the attacker went for the backups first →
The four routes of backup and cyber resilience — and which one is yours →
Backup vs replication vs snapshot vs archive
Four words sold by the same vendors and used as if they were tiers. They are not. Each answers a different failure mode — and a tool that handles one perfectly may be useless against another.
Backup
An independent copy, on different storage, kept for a retention period, restorable to a point in time. Answers: something was lost, corrupted or deleted — days or months ago — and you need that version back. Slow relative to the others; the only one that survives the loss of the source system and its storage.
Replication
A continuously updated second copy of a running system, ready to take over. Answers: the site or the server is down and you need to run elsewhere in minutes. Faithfully replicates corruption, deletion and ransomware — which is why it is the disaster-recovery guide, not this one.
Snapshot
A point-in-time image held on the same storage as the source (hypervisor or array). Answers: roll back a change made an hour ago. Fast and cheap — and gone with the storage it lives on. A snapshot is not a backup, and every backup vendor's first slide says so.
Archive
Data moved (not copied) to cheaper storage for retention and compliance, with an index. Answers: keep it for seven years and find it for the regulator. An archive is the only copy; it is not a recovery mechanism for the live system.
Six things decide this purchase. The feature grid is none of them.
Six variables decide this purchase. The instrument tests the ones documentation can verify (workloads, delivery form, instant recovery, hypervisor, meter); restore speed at scale and immutability mode are prose because the honest answers are a PoC figure and a configuration, not a datasheet line.
What it can protect
Physical, virtual, cloud, SaaS, containers, databases, NAS, endpoints — the list varies more than the brochures admit. Write yours down first, including the workload you forgot (the NAS, the cluster, the database on the old box).
Deployment model
Software you install (you design the repository), an appliance you rack (capacity is in the price), a service with nothing to run (restore speed is your WAN), or cloud-managed agents through a partner.
Restore granularity and speed at scale
Item-level vs full VM vs instant recovery that boots from the backup — and then the number nobody publishes: VMs per hour over your network. The real differentiator, and the one to test.
Licensing unit
Per VM or workload, per TB front-end, per TB back-end, per TB deduplicated, per socket, per device, per GiB-month. The same estate can cost 3× more or less on unit choice alone.
India data residency for backup copies
SaaS services with a documented Mumbai / Central India region (Druva, Commvault Cloud, Acronis), or your own storage where residency is your design; flagged where not documented.
Integration with the hypervisor and storage you have
VMware and Hyper-V are universal; Nutanix AHV and Proxmox are not; storage-array snapshot integration decides backup windows at scale.
Narrow 15 products to your shortlist
Set what is true for you. Products that fail a constraint fade with the reason on them; products we cannot verify for your case are marked and stay. Every chip is reversible.
Hypervisor
What it must protect
How it is delivered
Restore behaviour
The meter
India
Estate size
Restore speed at scale and immutability mode are in the notes below, not chips — the first is a PoC figure, the second is a configuration on most of these products and a default on few.
per workload (VUL) / year reported, by edition; list rose 4–8% in January 2026; storage is yours
The default for VMware and Hyper-V estates that want instant VM recovery, replication and a repository design they control — and now Nutanix AHV and Proxmox on the same licence.
The catch: Software you run: the repository, its hardening and the immutability mode are your design, not a default; Kubernetes is a separate product (Kasten) and SaaS is Veeam Data Cloud; no published list price.
per worker node / year; Enterprise edition; free edition for small clusters
Platform teams that need application-consistent backup, restore and mobility for Kubernetes namespaces — not a VM tool pointed at a cluster.
The catch: Kubernetes only — it protects nothing outside the cluster; priced per node on quote; immutability is whatever object-lock target you export to.
a Veeam Universal Licence per cloud instance, or per-instance marketplace billing; runs inside your cloud account
Estates on Veeam that also run EC2, RDS, Azure VMs or GCE and want native snapshots plus cost-tiered backup under the same console.
The catch: Cloud-native workloads only, one appliance per cloud; the object-lock bucket and its mode are yours to configure; not a replacement for the on-prem product.
per TB / month SaaS list (VMs and Kubernetes → databases); on-prem by front-end TB on quote; HyperScale X appliance; SaaS available in India
Mixed enterprises with the widest workload list — VMs, databases, NAS, Kubernetes, endpoints, cloud — that want one catalogue and a choice of software, appliance or SaaS.
The catch: Breadth is the product and the learning curve: a Commvault estate needs a Commvault administrator; the SaaS list is per TB and the on-prem price is a quote; the immutability mechanism depends on the target you choose.
per GiB-month of protected data, published; restores $0.08 per GiB plus AWS transfer; nothing to deploy
AWS-first estates that want EC2, EBS, RDS, S3 and DynamoDB backed up outside their own account, with a published per-GiB meter instead of a licence.
The catch: AWS only; the meter is gentle at rest and real at restore time (per GiB restored plus transfer); an India region for the vault is not documented.
per TB / year reported (licence, capacity-tiered); as-a-service $200–500+ per TB / year reported; appliance, software or BaaS
Enterprises that want scale-out appliances with instant mass restore and one platform that also runs cyber vaulting (FortKnox) and search over the backups.
The catch: Capacity-tiered reported pricing, no published list; the appliance form carries the platform's value — the software-only and BaaS forms are narrower; an India BaaS region is not documented.
per front-end TB / year reported at enterprise tiers, declining with capacity; Flex appliances separate
Large, long-lived estates — mainframe to cloud — that already run NetBackup and want its workload list and operating model to continue under Cohesity.
The catch: Enterprise-priced per front-end TB on quote; the heaviest operating model on this page; Proxmox support is not documented.
per back-end TB / month reported (premium support, prepaid); three-year minimums typical; appliance, partner hardware or cloud cluster
Enterprises that want backup and cyber recovery as one posture — append-only storage, anomaly detection and threat containment on the same platform.
The catch: Three-year, capacity-reviewed subscriptions with reported list around $130 per back-end TB per month — the most expensive meter here; an India data region is not documented.
per TB / month after deduplication (Business · Enterprise · Elite); AWS Mumbai region; built in Pune — nothing to run
Teams that want data-centre backup with no backup server, no repository and no patching — a SaaS service with an India region and an India engineering base.
The catch: No instant VM recovery from the cloud (restores and DRaaS instead); priced per deduplicated TB on quote; your data lives only in Druva's AWS — no on-prem copy.
per protected resource / month; snapshots stay in your cloud account, air-gapped copies in Druva's
AWS and Azure estates that want snapshot orchestration, cross-account copies and Data Lock from the same SaaS console as the data centre.
The catch: Native snapshots live in your account — your cloud root credential is the ceiling of their protection; cloud workloads only; quote-only.
per user / month reported (legacy inSync list; now Business · Enterprise · Elite per user); AWS Mumbai region
Laptop fleets that need silent, deduplicated backup with legal hold, eDiscovery and remote wipe — the endpoint half of Druva.
The catch: Endpoints only — servers, VMs and SaaS are other Druva SKUs; reported pricing, not published; no on-prem copy.
per workload / month reported (~$300 / year) or per GB, through partners; Acronis Cloud storage priced apart; Mumbai data centre
SMBs and MSP-run estates that want backup, anti-malware and patching in one agent, an India data centre and a per-workload or per-GB meter.
The catch: Sold and billed through partners — no public rate card; petabyte-scale estates are not where it is documented; immutability in Acronis Cloud defaults to governance mode (an admin can lift it) unless compliance mode is switched on.
per workstation / year list (Standard); servers and virtual hosts tiered higher; your storage
Organisations that want the Acronis agent and console on their own hardware — image backup, bare-metal restore and anti-malware without a partner in the loop.
The catch: Your storage, your immutability target; a per-seat list that climbs quickly on servers and hosts; not documented at petabyte scale.
per socket or server / year list (Vx virtual appliance); hardware appliances from $17,698 list; replication to Barracuda Cloud $799 per TB / year
Mid-market sites that want a backup appliance with LiveBoot recovery and a flat cloud replication price, sold through the channel.
The catch: VMware and Hyper-V only on the hypervisor side; an India region for Barracuda Cloud backup storage is not documented; the appliance's local copies are not the immutable ones — the cloud copies are.
per device / month reported, as an add-on to the RMM; image, file and folder backup for Windows, macOS and servers; Ninja cloud storage priced apart
IT teams already on NinjaOne RMM that want endpoint and server backup from the same agent and console, with cloud copies locked by S3 Object Lock.
The catch: No hypervisor-level VM backup, no databases, no NAS — it protects the devices the agent sits on; the local copy is not immutable, only the Ninja cloud copy; an India storage region is not documented.
Why each constraint rules out what it doesShow the reasoning ↓
Nutanix AHVRules out Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync) and NinjaOne Backup — does not protect virtual machines; Acronis Cyber Protect (on-prem) and Barracuda Backup (appliance · Vx · cloud) — Nutanix AHV not documented. That leaves Veeam Data Platform (Foundation · Advanced · Premium), Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Druva Hybrid Workloads and Acronis Cyber Protect Cloud (backup).
Proxmox VERules out Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync) and NinjaOne Backup — does not protect virtual machines; Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Druva Hybrid Workloads and Barracuda Backup (appliance · Vx · cloud) — Proxmox VE not documented. That leaves Veeam Data Platform (Foundation · Advanced · Premium), Commvault Cloud Backup & Recovery, Acronis Cyber Protect Cloud (backup) and Acronis Cyber Protect (on-prem).
Virtual machinesRules out Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync) and NinjaOne Backup — does not protect virtual machines. That leaves Veeam Data Platform (Foundation · Advanced · Premium), Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Druva Hybrid Workloads, Acronis Cyber Protect Cloud (backup), Acronis Cyber Protect (on-prem) and Barracuda Backup (appliance · Vx · cloud).
Public-cloud workloadsRules out Veeam Kasten for Kubernetes, Druva Hybrid Workloads, Druva Endpoints (inSync), Acronis Cyber Protect (on-prem), Barracuda Backup (appliance · Vx · cloud) and NinjaOne Backup — does not protect public-cloud workloads. That leaves Veeam Data Platform (Foundation · Advanced · Premium), Veeam Backup for AWS · Azure · Google Cloud, Commvault Cloud Backup & Recovery, Commvault Clumio (AWS), Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Druva Cloud Workloads (AWS · Azure) and Acronis Cyber Protect Cloud (backup).
DatabasesRules out Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync) and NinjaOne Backup — does not protect databases with application consistency. That leaves Veeam Data Platform (Foundation · Advanced · Premium), Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Druva Hybrid Workloads, Acronis Cyber Protect Cloud (backup), Acronis Cyber Protect (on-prem) and Barracuda Backup (appliance · Vx · cloud).
NAS and file sharesRules out Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync), Acronis Cyber Protect Cloud (backup), Acronis Cyber Protect (on-prem) and NinjaOne Backup — does not protect NAS / file shares. That leaves Veeam Data Platform (Foundation · Advanced · Premium), Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Druva Hybrid Workloads and Barracuda Backup (appliance · Vx · cloud).
KubernetesRules out Veeam Data Platform (Foundation · Advanced · Premium), Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Hybrid Workloads, Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync), Acronis Cyber Protect Cloud (backup), Acronis Cyber Protect (on-prem), Barracuda Backup (appliance · Vx · cloud) and NinjaOne Backup — does not protect Kubernetes. That leaves Veeam Kasten for Kubernetes, Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity) and Rubrik Security Cloud — Enterprise Edition.
Laptops and desktopsRules out Veeam Data Platform (Foundation · Advanced · Premium), Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Druva Hybrid Workloads, Druva Cloud Workloads (AWS · Azure) and Barracuda Backup (appliance · Vx · cloud) — does not protect endpoints. That leaves Commvault Cloud Backup & Recovery, Druva Endpoints (inSync), Acronis Cyber Protect Cloud (backup), Acronis Cyber Protect (on-prem) and NinjaOne Backup.
Backup as a serviceRules out Veeam Data Platform (Foundation · Advanced · Premium), Veeam Kasten for Kubernetes, Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Acronis Cyber Protect (on-prem) and Barracuda Backup (appliance · Vx · cloud) — you run the backup server, repository or appliance. That leaves Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Hybrid Workloads, Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync), Acronis Cyber Protect Cloud (backup) and NinjaOne Backup.
An applianceRules out Veeam Data Platform (Foundation · Advanced · Premium), Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Hybrid Workloads, Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync), Acronis Cyber Protect Cloud (backup), Acronis Cyber Protect (on-prem) and NinjaOne Backup — software or a service, not an appliance. That leaves Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition and Barracuda Backup (appliance · Vx · cloud). It flags Commvault Cloud Backup & Recovery — Software first; also sold as an appliance (HyperScale X / Flex) and NetBackup (Cohesity) — Software first; also sold as an appliance (HyperScale X / Flex) — marked on the cards, not removed.
Software on your hardwareRules out Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Cohesity DataProtect, Rubrik Security Cloud — Enterprise Edition, Druva Hybrid Workloads, Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync), Acronis Cyber Protect Cloud (backup), Barracuda Backup (appliance · Vx · cloud) and NinjaOne Backup — an appliance or a service, not software you install. That leaves Veeam Data Platform (Foundation · Advanced · Premium), Veeam Kasten for Kubernetes, Commvault Cloud Backup & Recovery, NetBackup (Cohesity) and Acronis Cyber Protect (on-prem).
Instant recoveryRules out Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Hybrid Workloads, Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync) and NinjaOne Backup — no documented instant recovery; restores copy data back first. That leaves Veeam Data Platform (Foundation · Advanced · Premium), Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition, Acronis Cyber Protect Cloud (backup), Acronis Cyber Protect (on-prem) and Barracuda Backup (appliance · Vx · cloud).
Per workload, not per TBRules out Commvault Cloud Backup & Recovery, Cohesity DataProtect, NetBackup (Cohesity), Rubrik Security Cloud — Enterprise Edition and Druva Hybrid Workloads — priced per TB (front-end or deduplicated). That leaves Veeam Data Platform (Foundation · Advanced · Premium), Veeam Kasten for Kubernetes, Veeam Backup for AWS · Azure · Google Cloud, Commvault Clumio (AWS), Druva Cloud Workloads (AWS · Azure), Druva Endpoints (inSync), Acronis Cyber Protect Cloud (backup), Acronis Cyber Protect (on-prem), Barracuda Backup (appliance · Vx · cloud) and NinjaOne Backup.
Copies in IndiaRules nothing out on published terms. It flags Veeam Data Platform (Foundation · Advanced · Premium) — Your storage, Veeam Kasten for Kubernetes — Your storage, Veeam Backup for AWS · Azure · Google Cloud — Your storage, Commvault Clumio (AWS) — India region for the service's storage not documented, Cohesity DataProtect — India region for the service's storage not documented, NetBackup (Cohesity) — Your storage, Rubrik Security Cloud — Enterprise Edition — India region for the service's storage not documented, Acronis Cyber Protect (on-prem) — Your storage, Barracuda Backup (appliance · Vx · cloud) — India region for the service's storage not documented and NinjaOne Backup — India region for the service's storage not documented — marked on the cards, not removed.
Above a petabyteRules nothing out on published terms. It flags Acronis Cyber Protect Cloud (backup) — Unverified at petabyte scale, Acronis Cyber Protect (on-prem) — Unverified at petabyte scale, Barracuda Backup (appliance · Vx · cloud) — Unverified at petabyte scale and NinjaOne Backup — Unverified at petabyte scale — marked on the cards, not removed.
Restore speed at scaleInstant recovery is documented for Veeam, Commvault, Cohesity, NetBackup, Rubrik, Acronis and Barracuda — booting or mounting from the backup. What no datasheet documents is the number that matters: how many VMs per hour come back over your network from your repository. That is a PoC figure; TechBag's delivery figures per platform are [TechBag to confirm].
Where immutability is enforcedThree patterns on this page. Target-enforced (Veeam, Commvault, NetBackup, Acronis on-prem, Kasten, Veeam public cloud): the product writes to object lock, a hardened repository or WORM hardware that you configure — as strong as your configuration. Vendor-controlled (Druva, Clumio, Acronis Cloud, Barracuda Cloud, NinjaOne cloud): the service's storage, locked by the vendor — read the mode (governance vs compliance). Software-enforced on the platform (Cohesity DataLock, Rubrik's append-only filesystem with Retention Lock). The cyber-recovery guide tabulates this per SKU; here it is a note because every product on this page can be made immutable — and most are not, by default.
Licensing unitsPer workload (Veeam VUL, Acronis), per TB front-end (Commvault on-prem, NetBackup), per TB back-end per month (Rubrik), per TB after deduplication (Druva), per socket (Barracuda Vx), per device (NinjaOne), per GiB-month (Clumio), per node (Kasten). The same estate can cost 3× more or less depending only on which unit the vendor chose — model your own VM count, front-end TB and growth before comparing quotes.
Storage is not in the licenceEvery line on this page excludes the storage the backups sit on — appliance capacity, object storage, cloud tiers, a second copy off-site. It is frequently larger than the licence (see the cost section). Druva, Clumio and Acronis Cloud bundle or meter storage differently; ask which.
Under 10 TBRules nothing out on documentation: Acronis, NinjaOne, Barracuda Vx and Veeam Community / Foundation are sold to small estates; Commvault, Cohesity, Rubrik and NetBackup are enterprise-positioned but publish no floor. Which ones are a poor fit below 10 TB is delivery judgement: [TechBag to confirm].
Eight situations, eight shortlists — with the restore named
Each shortlist names what the restore will look like in that estate, not only which products back it up. If your hypervisor is changing, start from that row.
VMware or Hyper-V, 50–500 VMs, one or two sites — the common case
Why: Veeam is the default for this estate — instant VM recovery, replication and a repository you design; Commvault Cloud covers the same VMs with SaaS or software and a wider catalogue; Acronis brings the agent with backup and security in one, through a partner, with a Mumbai data centre.
The trade-off: Veeam makes you the storage architect; Commvault makes you learn Commvault; Acronis makes you pick a partner.
Leaving VMware for Nutanix AHV or Proxmox
Why: Veeam and Commvault document both AHV and Proxmox; Cohesity documents AHV. The backup product is the one thing that should not also change mid-migration.
The trade-off: Proxmox support is newer everywhere — test the restore, not the backup, on the new hypervisor before the cut-over.
Nothing to run — no backup server, no repository, no patching
Why: Druva is SaaS-only with an AWS Mumbai region and a Pune engineering base; Commvault Cloud's SaaS form is available in India; Acronis runs from its Mumbai data centre through partners.
The trade-off: No local copy means restore speed is your internet line — size the WAN before the contract; instant VM recovery is not what SaaS backup does.
An enterprise appliance with instant mass restore
Why: Cohesity's scale-out appliances document instant mass restore; Rubrik's appliances document Live Mount on append-only storage; Barracuda's appliances document LiveBoot with a flat cloud price for the smaller estate.
The trade-off: Appliances bundle capacity into the price — you buy the next shelf, not the next TB; three-year terms (Rubrik) and capacity tiers (Cohesity) are the real negotiation.
AWS-heavy estate — EC2, RDS, S3 are most of the data
Why: Clumio backs up outside your account on a published per-GiB meter; Veeam Backup for AWS runs inside your account on VUL or marketplace billing; Druva orchestrates snapshots and air-gapped copies from the same console as the data centre.
The trade-off: Native snapshots in your own account are only as safe as your root credential; an outside-the-account copy costs at restore time (per GiB and transfer).
Kubernetes is production now
Why: Kasten is built for namespaces and application consistency; Commvault and Cohesity protect Kubernetes inside the same platform as the VMs.
The trade-off: A Kubernetes-native tool is a second backup product to run; the platform route is one product with shallower Kubernetes depth — decide which team owns the restore.
MSP-run SMB — laptops, a few servers, no backup administrator
Why: Acronis and NinjaOne are the agent-based, RMM-adjacent options priced per workload or device through partners; Barracuda Vx puts an appliance in the rack at $599 per socket per year list.
The trade-off: Device-level backup does not protect the hypervisor layer; if a host dies, you rebuild and restore guests one by one.
Large estate already on NetBackup or Commvault
Why: NetBackup continues under Cohesity with its workload list intact; Commvault is the like-for-like alternative with the widest catalogue; Rubrik is the posture-led replacement when cyber recovery is the reason to move.
The trade-off: Switching at this size is a retention-tail project: the old product stays read-only until its last backup expires. Price the overlap.
The restore ladder, and why the meter decides the bill
Every product here restores. They differ in how much comes back per hour and what you pay per unit of what. Place each product on both before comparing quotes.
Rung 1
Item-level
One file, one mailbox, one table row — from a catalogue or an indexed backup. Everything on this page does it; the difference is how long the index takes to search at scale.
Rung 2
Full workload
A whole VM, server or database copied back to production storage. Speed is your network and your repository's read throughput — the number to measure, not to read.
Rung 3
Instant recovery
The VM boots from the backup repository in minutes while data migrates behind it. Documented at Veeam, Commvault, Cohesity, NetBackup, Rubrik, Acronis and Barracuda.
Rung 4
Mass restore
Hundreds of VMs at once, from scale-out storage, with orchestration — the ransomware-day requirement. Cohesity and Rubrik document it on their appliances; NetBackup and Commvault at enterprise scale.
The meter test
Model your estate on every unit before you read a quote.
- —Per workload (Veeam VUL, Acronis) is cheap for large VMs with lots of data and expensive for hundreds of small ones.
- —Per front-end TB (Commvault on-prem, NetBackup) is the reverse — and grows with your data whether or not the VM count does.
- —Per back-end TB per month (Rubrik) and per deduplicated TB (Druva) reward data that compresses; test your own dedupe ratio, not the brochure’s.
- —Per socket / device / GiB-month (Barracuda Vx, NinjaOne, Clumio) are honest at small scale and surprising at large — run the growth curve to year three.
Storage, separately
The licence is rarely the largest number.
- —Object storage for a second copy: Veeam Data Cloud Vault $14–24 per TB per month; hyperscaler hot tiers ~$20 and cool tiers ~$10 per TB per month; S3-compatible providers from ~$7. At 100 TB that is $16800 ≈ ₹13,94,400–$28800 ≈ ₹23,90,400 a year before a single licence.
- —Appliance capacity (Cohesity, Rubrik, Barracuda) is in the hardware price — you buy the next shelf, not the next TB.
- —SaaS services (Druva, Clumio, Acronis Cloud) bundle or meter storage differently — ask whether the TB you are quoted is licence, storage, or both.
- —Egress at restore is the line everyone forgets until the day: Clumio publishes $0.08 per GiB restored; hyperscalers charge transfer out.
What changes at 10 TB, 100 TB and a petabyte
Backup scales by data volume and by the number of things to protect — and the two meters diverge exactly here. The bill follows the unit; the backup window follows the network.
The agent and the admin are the constraint
- —Any product here protects it; the question is who runs it — Acronis, NinjaOne and Barracuda Vx are built for the estate with no backup administrator.
- —A single repository or a SaaS target is enough; instant recovery is a convenience, not a requirement.
- —Per-device and per-workload meters are cheapest at this size.
Put this in your PoC
Restore one full VM to a different host and time it. If nobody has ever done it, that is the finding.
The repository and the meter are the constraint
- —Backup windows start to depend on storage-snapshot integration and changed-block tracking — VMware and Hyper-V are universal, AHV and Proxmox are not.
- —The licence unit flips: per-TB products overtake per-workload ones if VMs are large; model both.
- —A second, immutable copy off-site becomes a line item larger than the licence — object storage at $14–24 per TB per month.
Put this in your PoC
Ask for the dedupe ratio on your data, not the datasheet's, and the VMs-per-hour restore figure from a reference at your size.
Mass restore and the operating model are the constraint
- —Scale-out appliances (Cohesity, Rubrik) and enterprise software (NetBackup, Commvault) document petabyte estates; Acronis, NinjaOne, Barracuda and Druva Endpoints are flagged unverified here, not ruled out.
- —Restore orchestration — boot order, dependencies, hundreds of VMs — is the difference between a bad week and a bad quarter.
- —Capacity-tiered subscriptions and three-year terms are the negotiation; the list price is a starting point nobody pays.
Put this in your PoC
Run a mass-restore rehearsal of 50 VMs from the immutable copy. The hours it takes is the only number the board should hear.
Veeam, Commvault, Cohesity, NetBackup, Rubrik, Clumio and Druva’s data-centre products document petabyte estates; Acronis, NinjaOne, Barracuda and Druva Endpoints are flagged unverified at that size. Where a specific product strains for your estate: [TechBag to confirm].
Leaving a backup product is a retention-tail project
Backup formats are proprietary. Switching does not migrate the old backups — it leaves them behind, readable only by the old product, until the last one expires.
The retention tail
Every backup taken by the old product stays in its format. Keep the old product licensed (or at least installed and read-only) until the longest retention expires — seven years for some regulated data.
Re-seeding
The new product takes a full first backup of everything: the network and the repository feel it. Plan a seeding window; for SaaS targets, plan the WAN.
Agents and plugins
Application agents (databases, Exchange, SAP) are per product and must be replaced and re-tested for consistency; hypervisor integration is re-done per host.
The immutable copy
If the old product's immutable copy was the ransomware plan, it stops being refreshed on day one of the switch. The new product's immutable copy must exist before the old one is the only one.
Retention tail, seeding weeks and overlap cost for your estate: [TechBag to confirm] — TechBag scopes it from your retention policies and data volume.
Per workload or per TB — and then the storage
What you may already hold, the products priced on their own meters at three estate sizes in USD and INR, and what the licence line leaves out — which, for backup, starts with the storage the backups sit on.
Do you already own one?
Four places a backup may already seem to exist. Only one is.
- Hypervisor-native snapshots
- No A snapshot lives on the same storage as the VM and dies with it. Useful for rollback; not a backup, not retention, not a restore from another site.
- Storage-array replication
- No A second array that faithfully replicates corruption, deletion and encryption. That is disaster recovery’s half, not backup’s — and only if the array is elsewhere.
- Your cloud provider's backup service
- Partly AWS Backup and Azure Backup are real backup for their own cloud, with vault lock / immutable vault. They do not reach your data centre or each other; a multi-cloud estate needs one catalogue.
- Backup inside the RMM you run
- Partly NinjaOne Backup and Acronis through an MSP protect the devices the agent sits on — servers and laptops — not the hypervisor, the database or the NAS.
If a copy you already pay for is enough for your failure modes, we say so. It costs us a sale and saves you one.
What the rest actually cost
Published and reported meters (INR for scale), then worked at three estate sizes — each product on its own unit, so the grid shows what the unit does to the bill. Storage is a separate block below it, on purpose.
- Veeam Data Platform(reported $250–450 per workload / yr)$12,500–22,500 ≈₹10,37,500–₹18,67,500
- Commvault Cloud SaaS(list $58.50–90 per TB / mo)$7,020–10,800 ≈₹5,82,660–₹8,96,400
- Cohesity DataProtect(reported $150–400 per TB / yr)$1,500–4,000 ≈₹1,24,500–₹3,32,000
- NetBackup(reported $400–900 per front-end TB / yr)$4,000–9,000 ≈₹3,32,000–₹7,47,000
- Rubrik Enterprise Edition(reported ~$130 per back-end TB / mo; ~1× assumed)$15,600 ≈₹12,94,800
- Acronis Cyber Protect Cloud(reported ~$300 per workload / yr)$15,000 ≈₹12,45,000
- Clumio(published $0.012 per GiB-month, AWS data)$1,475 ≈₹1,22,425
- Barracuda cloud replication(list $799 per TB / yr; appliance apart)$7,990 ≈₹6,63,170
- Druva Hybrid Workloads(per deduped TB)Quote
- Veeam Kasten(per node)Quote
- NinjaOne Backup(per device)Quote
- Veeam Data Platform(reported $250–450 per workload / yr)$75,000–1,35,000 ≈₹62,25,000–₹1,12,05,000
- Commvault Cloud SaaS(list $58.50–90 per TB / mo)$70,200–1,08,000 ≈₹58,26,600–₹89,64,000
- Cohesity DataProtect(reported $150–400 per TB / yr)$15,000–40,000 ≈₹12,45,000–₹33,20,000
- NetBackup(reported $400–900 per front-end TB / yr)$40,000–90,000 ≈₹33,20,000–₹74,70,000
- Rubrik Enterprise Edition(reported ~$130 per back-end TB / mo; ~1× assumed)$1,56,000 ≈₹1,29,48,000
- Acronis Cyber Protect Cloud(reported ~$300 per workload / yr)$90,000 ≈₹74,70,000
- Clumio(published $0.012 per GiB-month, AWS data)$14,746 ≈₹12,23,918
- Barracuda cloud replication(list $799 per TB / yr; appliance apart)$79,900 ≈₹66,31,700
- Druva Hybrid Workloads(per deduped TB)Quote
- Veeam Kasten(per node)Quote
- NinjaOne Backup(per device)Quote
- Veeam Data Platform(reported $250–450 per workload / yr)$5,00,000–9,00,000 ≈₹4,15,00,000–₹7,47,00,000
- Commvault Cloud SaaS(list $58.50–90 per TB / mo)$7,02,000–10,80,000 ≈₹5,82,66,000–₹8,96,40,000
- Cohesity DataProtect(reported $150–400 per TB / yr)$1,50,000–4,00,000 ≈₹1,24,50,000–₹3,32,00,000
- NetBackup(reported $400–900 per front-end TB / yr)$4,00,000–9,00,000 ≈₹3,32,00,000–₹7,47,00,000
- Rubrik Enterprise Edition(reported ~$130 per back-end TB / mo; ~1× assumed)$15,60,000 ≈₹12,94,80,000
- Acronis Cyber Protect Cloud(reported ~$300 per workload / yr)$6,00,000 ≈₹4,98,00,000
- Clumio(published $0.012 per GiB-month, AWS data)$1,47,456 ≈₹1,22,38,848
- Barracuda cloud replication(list $799 per TB / yr; appliance apart)$7,99,000 ≈₹6,63,17,000
- Druva Hybrid Workloads(per deduped TB)Quote
- Veeam Kasten(per node)Quote
- NinjaOne Backup(per device)Quote
Storage — stated apart from the licence, on purpose
TechBag gives INR pricing, GST, PO cycle, minimums and tier-matched quotes. The INR above is conversion for scale at ≈₹83/$; the tier-matched INR quote is ours.
What isn’t in the licence price
The storage
Object storage, appliance capacity, a second copy off-site — see the block above. At 100 TB the second copy alone is $16800 ≈ ₹13,94,400–$28800 ≈ ₹23,90,400 a year at vault list prices, before the hardware under the first.
Egress and restore fees
Per-GiB restore charges (Clumio $0.08), hyperscaler transfer out, and the WAN upgrade a SaaS restore turns out to need. Price a full 10 TB restore, not a file.
Restore testing, in hours
A quarterly full-VM restore, an annual mass-restore rehearsal, and a person to own the results. Vendors automate verification (Veeam SureBackup, Commvault, Cohesity, Rubrik); the rehearsal is still yours. Your hours: [TechBag to confirm].
What goes wrong
Documented behaviour and cost curves, cross-checked against TechBag engagements before any becomes a named case. Each is cheaper to read here than to discover on the day.
Backups that succeed and restores that fail
Green dashboard every night; the first real restore hits a corrupt chain, a missing agent or an untested application-consistency step. Nobody had restored at volume. Rehearse it.
Per-TB pricing scaling in a way nobody modelled
Front-end TB grew 40% a year; the licence followed; the budget did not. Model the meter on your growth curve to year three before signing.
Agent coverage gaps on the one workload that mattered
The NAS, the old database server, the Proxmox cluster — outside the product's list or never licensed. Write the workload list down, including the one you forgot.
Restore times that meet the SLA on paper and not on the network
Instant recovery boots the VM; full performance waits on data migrating over a 1 Gbps link from a repository built for writes. Measure VMs per hour, not seconds to boot.
Backup windows that stop fitting
Full backups that used to finish by 6am now run into the working day — no storage-snapshot integration, no changed-block tracking on the new hypervisor. The window is a design, not a setting.
Storage larger than the licence, discovered at renewal
The licence was negotiated hard; the object storage, the second appliance shelf and the egress were not in the business case. They were most of the bill.
A snapshot mistaken for a backup
The array died, or the hypervisor datastore was encrypted — and every snapshot went with it. Snapshots roll back; they do not restore from elsewhere.
Immutability assumed, never enabled
The product supported object lock; the bucket was created without it. Support is not a default — the cyber-recovery guide is the checklist.
Three doors — pick by where you are
Vendor-neutral. No gated content.