Published tests Lab The tests we run before selling, with the date, the method and what we don’t know yet. No customer, server or domain names.
23 posts · updated Oct 19, 2026
Suggest a test 21 posts
3 cPanel servers measured
371 accounts observed in one night
All 23 Performance 11 Storage 2 Security 0 Compatibility 10
Performance Oct 19, 2026 250 s for the 6 accounts to come back, 3 at a time; JetBackup took 839 s in the same window Restoring WordPress with Salvo faster than with the cPanel package: 6 real sites, one and many accounts at a time We did not like the restore in Lab 23 and changed how Salvo restores. A 3.2 GB account came back 36% faster than the panel’s own package, and the 6 accounts in 250 s, with every message. Lab 24 Read · 6 min
Compatibility Oct 19, 2026 99.98% of customer files back, with every message WordPress backup with email: 6 real sites restored by Salvo and by JetBackup Six real accounts, the same storage, backup and restore by both tools. Salvo stored 21% less and brought every message back; JetBackup left 648 out and restored faster. Lab 23 Read · 5 min
Performance Oct 18, 2026 3.3 s to bring wp_options back, instead of 121 s WordPress: restore only the table that broke, in 3 seconds instead of 2 minutes Five real WordPress databases. wp_options, wp_postmeta and wp_posts restored on their own, without undoing the rest of the database: a big gain on the tables that break most often, and all of them came back identical. Lab 22 Read · 4 min
Performance Oct 19, 2026 5.2× faster after the improvements: 10 GB in 5 min 45 s We improved Salvo’s restore: a 10 GB account back in 5 min 45 s, at the storage’s worst hour We measured the restore of a real 10 GB account at the backup storage’s worst moment and did not like it: almost 30 minutes. We found the bottleneck, made two improvements to Salvo, and the same account came back in 5 min 45 s, identical. Lab 21 Read · 4 min
Compatibility Oct 17, 2026 5 of 5 accidents undone, checked item by item Restore only what you need: a database, a mailbox, a few files, all back exactly as they were A broken database, deleted messages, a deleted mailbox, overwritten files: each one restored on its own, without touching the rest, from the panel and the command line. Lab 20 Read · 3 min
Compatibility Oct 17, 2026 ~2 min from deleted account to account back The account is gone: back, whole, from the backup, in about 2 minutes A real account deleted from the server and restored straight from the backup on the storage, from the command line and the panel: all 23 thousand messages back both times. Lab 19 Read · 3 min
Compatibility Oct 17, 2026 100% of messages identical Restoring a cPanel backup through Salvo: four times, the same result, and 100% of messages identical One real account restored four times on another server, from the panel and the command line: the same content all four times, and what the server felt during the restore. Lab 18 Read · 3 min
Compatibility Oct 16, 2026 99.96% of customer files identical Disaster recovery: 10 whole accounts on another server, 99.96% of customer files identical With the source server out: the 10 accounts came back on an older cPanel. 0 emails lost, 987 tables identical to the backup, and what came back different, explained. Lab 16 Read · 4 min
Performance Oct 16, 2026 152× fewer storage reads The incremental right after: 152× fewer reads on the storage disk Same 10 accounts as Lab 14, right after the first backup: 155 reads on the storage disk on Salvo, versus 23,580 on JetBackup. Lab 15 Read · 3 min
Compatibility Oct 16, 2026 166× fewer files on the storage The first backup side by side: 45% less time and 166× fewer files on the storage 10 accounts, the same rules, each on its own storage volume measured every 5 s. Includes where Salvo loses: CPU, load and memory on the source server. Lab 14 Read · 3 min
Compatibility Oct 16, 2026 7.7× less total time The next-day backup on the same 6 accounts: 494 s on JetBackup, 64 s on Salvo Same 6 accounts, same server, next-day backup. The conditions weren’t the same (time of day and accounts in parallel), and the post shows which. Lab 13 Read · 3 min
Performance Oct 16, 2026 4 s for 21 thousand files, cold The next-day backup, before dawn and cold: 4 s for 21 thousand files With the engine’s cache empty: 301 files read and 21,019 skipped without opening. The scheduled account had been migrated, and Salvo marked the failure with the reason. Lab 12 Read · 2 min
Compatibility Oct 15, 2026 20 s to restore the whole account The whole account restored on another server: domains, email, identical database and grants A Salvo backup became a cPanel package and was restored on another server: domains, mailboxes, 15 tables with identical rows, grants and cron. Lab 11 Read · 2 min
Storage Oct 15, 2026 −35% data read What doesn’t need a backup: 35% less read, 31% less stored With the default exclusion list (caches, logs, plugin backups), the first backup of 6 accounts read 35% less and wrote 31% less. Email stays off the list. Lab 10 Read · 2 min
Compatibility Oct 15, 2026 33 s 3 full cPanel accounts 3 full cPanel accounts in 33 s on a production server; the incremental, in 7 s Configuration through pkgacct, files, email and databases, on a production server. The test caught databases with @ in the name and marked them as failed. Lab 09 Read · 3 min
Performance Oct 15, 2026 141 MB/s uploaded by the engine The engine uploads 141 MB/s reading one file per core; 4 accounts together used 2.4 GB of memory Reading one file at a time, 60 MB/s; one per core, 141 MB/s. The first backup of 4 accounts together took 29 s, with 2.4 GB of peak memory. Lab 08 Read · 2 min
Performance Oct 15, 2026 8 ms per MB for compression Compression is the costliest step of the backup; encryption barely registers We measured hash, compression and encryption on the processor alone. Compression costs 8 ms per MB on one core; encryption, 1.1 ms. Lab 07 Read · 2 min
Performance Oct 15, 2026 3× faster with aes128-gcm The limit was the encryption, not the network: switching ciphers tripled the upload One SSH connection went from 78 to 245 MB/s just by switching ciphers; with 4 connections, 867 MB/s. Salvo already uses that cipher. Lab 06 Read · 2 min
Performance Oct 15, 2026 ~90 MB/s per SSH connection Writing to the storage disk was as fast as throwing the data away We measured the upload to the storage piece by piece: network, encryption and disk. The disk is not the limit; the ceiling is ~90 MB/s per connection. Lab 05 Read · 2 min
Storage Oct 15, 2026 ½ the space Compression cuts space in half; keeping each account separate costs 28% One repository per account keeps accounts independent, but loses files repeated across them. We measured the price. Lab 02 Read · 3 min
Performance Oct 14, 2026 30 MB of cache 30 MB of cache for 233 thousand files, and deleting it cost nothing With the cache deleted, the incremental took 18 s, versus 19 s with a full cache. With no cache at all, 4× slower. Lab 01 Read · 2 min
Performance Oct 14, 2026 1 hardlink per file, per night One new hardlink per file, every night: why backup storage saturates With rsync and hardlink incrementals, even unchanged files cause random writes on the storage. Mechanism and estimate. Lab 04 Read · 2 min
Compatibility Oct 13, 2026 87% transferring 371 accounts in 2h10: how JetBackup 5 works through a backup night We watched, without changing anything, one backup night on a cPanel server with about 370 accounts. Lab 03 Read · 3 min
In progress The chart goes here once the test ends and the post is approved.
The engine’s limit: SFTP library or storage write Chart awaiting data
Suggest a test What do you want to see tested? The Lab’s next tests also come from the questions of people who run servers. Tell us what you want measured and how it would help your operation.
How we choose What shows up in several suggestions weighs more. It has to be measurable on a test server, with no customer data. We turn the result into a post, with method and caveats.