How cloud DR scores work on this site

What question do the scores answer?

One buyer question: after an outage, a bad change or the loss of a region, which tool gets your cloud infrastructure back, not only your data, fastest and most completely. A tool can be excellent at something else, such as long-term data retention or infrastructure-as-code governance, and still score low here.

Why these weights?

Rebuild scope carries the most weight (20%) because completeness is half of the question: a recovery that returns the data into an environment that no longer exists does not return the service. Recovery orchestration, region and account loss, and bad-change rollback get 15% each because they map to the three incidents the question names: orchestration decides how long an outage lasts, a region or account loss needs a documented second target, and a bad change needs a known-good point to return to. Published recovery targets and recovery testing get 10% each because they are the evidence for the word fastest: a stated RTO or RPO is something to hold a vendor to, and a recovery that has been rehearsed is faster than one that has not. Data inside the rebuild gets 10% because infrastructure without its data is not a recovery, but most teams already own a data backup product. Pricing transparency gets 5%: it does not change how fast you recover, but a price you cannot see is a price you cannot compare. Restore as code is not scored separately; profiles state each tool's position on it.

How is a score decided?

• 9 to 10: documented in detail in public docs or pricing pages and central to the product.
• 7 to 8.9: documented, with gaps in scope or detail.
• 4 to 6.9: partly present, present only in one tier or for some workloads, or described only at marketing level.
• 1 to 3.9: minimal or indirect.
• 0 to 0.9: absent.
• Where a vendor does not publish something, we score what is published and say so in the reason. Missing information lowers a score; we do not guess in either direction.

How is the total calculated?

Total = (score on rebuild scope x 20 + score on recovery orchestration x 15 + score on region and account loss x 15 + score on bad-change rollback x 15 + score on published recovery targets x 10 + score on recovery testing x 10 + score on data inside the rebuild x 10 + score on pricing transparency x 5) / 100. The code computes the exact total; pages show it to one decimal.

What do the recovery bricks show?

Each tool gets three bricks, one per incident in the question: an outage (a service or zone fails and the application must come back), a bad change (a misconfiguration, deletion or faulty automated change must be undone) and a region loss (a whole region or account is lost and the environment must be rebuilt elsewhere). A solid brick means the vendor documents recovery from that incident, a hatched brick means it is partly documented, and a dashed outline means it is not documented on the pages reviewed.

Bricks are not scored separately. They summarize what each vendor's pages say about the three incidents in the question.

What counts as evidence?

• Only pages we read: vendor sites, documentation, pricing pages, announcements, standards bodies and reference pages for company history.
• Every vendor figure, such as an RTO, RPO or setup time, is quoted as the vendor's claim with a link, never as our finding.
• Every fact shows the date we read its source. In this edition vendor pages were read on 27 September 2026 or 11 October 2026.
• Where a fact is not on the pages reviewed, we write Not published or Not documented on the pages reviewed.
• We do not use review-site ratings, analyst placements we could not read, or unnamed customer stories.

Which tools are scored, and which are not?

Scored on all eight criteria, in rank order: Arpio, Firefly, Commvault Cloud Rewind, ControlMonkey and Gambit Security.

• Data backup: works alongside, not scored: They protect workload data rather than rebuilding cloud infrastructure, so they complement the tools we score. Profiles: Cohesity.
• IaC management: outside the disaster recovery question, not scored: They run and govern infrastructure as code but describe no backup or cloud disaster recovery product. Profiles: StackGuardian, Scalr.

What changed in this edition?

• 2026-10-13: ControlMonkey scored on all eight criteria; Cohesity, Scalr and StackGuardian moved to unscored groups and their scores removed, not changed.

What are the limitations?

• Public sources only: no product access, no hands-on testing, no vendor interviews or briefings.
• No score reflects a restore we ran or a recovery time we measured.
• Point in time: facts can change after the read date shown next to each source; check the vendor's current pages before buying.
• The weights express one editorial view of this question. The recovery time estimator lets you change them.

How do corrections work?

Send a correction through the contact page with the topic Correction and a link to the public source. We check it against the vendor's page and update the data; totals recompute automatically and the change is dated on the page.