Most people judge online backup services by price pages and review scores. That tells you nothing about the day a file is encrypted or deleted. Backblaze reported that Computer Backup revenue slipped 2% to US$16.2 million in Q3 2026 while B2 Cloud Storage grew 47%, so the consumer backup category is quiet, not dead, and you still need to pick a service on evidence rather than marketing.Why Test Online Backup Services Against RansomwareWhat You Are Testing and What It CostsPrerequisites and VersionsSteps 1 to 3: Build the Lab Dataset and Hash ManifestStep 1: Create the lab folder and generate test filesStep 2: Hash every file into a baseline manifestStep 3: Prepare the timing logSteps 4 to 6: Install the Clients and Take the Clean BaselineStep 4: Backblaze Computer BackupStep 5: IDrive PersonalStep 6: Wasabi plus KopiaSteps 7 and 8: Run a Safe Ransomware SimulationStep 7: Run the attack script on the lab folderStep 8: Let each service back up the poisoned stateSteps 9 to 11: Restore the Clean Copy and Verify Every FileStep 9: Restore from Backblaze and IDriveStep 10: Restore from Wasabi with KopiaStep 11: Verify every restored file against the manifestSteps 12 and 13: Score the Results and Model Five-Year CostStep 12: Fill in the scorecardStep 13: Model five-year cost for your real data sizeAustralian Considerations: NBN Upload, Data Location and PrivacyCommon Pitfalls That Ruin the TestWhat Good Results Look Like: Sample OutputTroubleshooting: 10 Problems and FixesAdvanced Tips for a Sharper Backup TestThrottle uploads so the PC stays usableRepeat the test with a delayed discoveryAdd a second, independent copyTest a single-file restore as a separate drillComplete Working Project: One Script to Report the Whole Bake-OffFAQ: Online Backup ServicesWhat are the best online backup services for a single Windows PC in Australia?How long does the first online backup take?Is online backup safe against ransomware?Does Backblaze Computer Backup cover external drives and NAS devices?Why use Wasabi with a separate client instead of a ready-made backup app?Will the prices in this guide change?Can I keep my backup data in Australia?Related Coverage
This tutorial gives you that evidence in about 90 minutes. You will build a throwaway lab folder on a Windows PC, back it up to three online backup services, run a safe ransomware simulation against it, and then restore the clean copy. Every restore gets checked file by file against a SHA-256 manifest, so you end with hard numbers: how many files came back intact, how long recovery took, and what the whole setup costs over five years. If you only want a ranking without the hands-on work, our roundup of the best cloud backup services for Australians covers that ground.
The method is deliberately boring. Same data, same PC, same internet connection, same attack, same checks. When the variables are locked down, differences between online backup services show up fast. One service might keep old versions of changed files while another quietly overwrites them with the encrypted copies. A third might restore everything but take three times as long. You will not see that on any pricing table.
Google · Preferred Sources
Don’t miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
Add Now
Why Test Online Backup Services Against Ransomware
A backup that cannot undo damage is only a copy. The failure mode that hurts most is not hardware dying. It is a sync-style backup faithfully uploading corrupted files overnight, then pruning the good versions after a retention window you never checked. Backblaze itself makes this point for Computer Backup with Enterprise Control, which it describes as providing one year of version history, Legal Hold and IT-managed restores from an admin console.
For a home user or a small office in Australia, the practical questions are simpler. Can you step back to a version from before the attack? Does the service keep deleted files long enough for you to notice? How big a restore can you request in one go? Backblaze, for example, lists zip restores of up to 500GB per request on its own page, which matters when you are rebuilding a whole drive rather than a single document.
You may like
There is also a structural reason to test rather than trust. The three options in this guide are three different kinds of product. Backblaze Computer Backup and IDrive Personal are agent-based services where software on your PC decides what to send. Wasabi Hot Cloud Storage is raw S3-compatible object storage, so you bring your own backup client. Comparing them fairly means running the same recovery scenario through each model and seeing where each one bends.
What You Are Testing and What It Costs
The bake-off covers two agent-based services and one do-it-yourself object-storage setup. Prices below come from each company’s own pages as fetched on November 8, 2026, and they are in US dollars. Convert at your own bank’s rate, because exchange movement will shift the Australian dollar figure by the time you read this.
ServiceTypeListed price (USD)Role in this testBackblaze Computer BackupAgent, unlimited data for one computer$9/month, or $189 on a two-year planSubject 1IDrive PersonalAgent, 5TB, multiple computers$9.99/month standard, $6.99/month billed annuallySubject 2Wasabi Hot Cloud Storage plus KopiaS3-compatible storage plus open-source client$7.99/TB/month, 1TB monthly minimum, 90-day minimum storageSubject 3Backblaze B2 plus KopiaS3-compatible storage plus open-source client$6.95/TB/month, first 10GB freeOptional subject 4IDrive e2 plus KopiaS3-compatible storage plus open-source clientFrom $59.50/year for 1TB ($29.75 first year)Optional subject 5
Read the price column carefully. IDrive’s lower figures are promotional or annual-billing prices, and the first-year promotions are not renewal prices. Backblaze’s Computer Backup subscription and its B2 storage price are different products, so never substitute one for the other when you model costs. Reports put the Computer Backup annual price at US$99, but the retrieved official page showed the monthly and two-year figures, so treat the annual number as a secondary-source figure and confirm it at checkout.
The whole test fits inside roughly US$27 at monthly list prices if you pay Backblaze $9, IDrive $9.99 and Wasabi $7.99 for one month. Wasabi advertises a 30-day free trial on its sales page, which may bring that figure down, though the offer can change. Cancel anything you do not keep before the next billing date. Wasabi’s 90-day minimum storage rule means data you delete early is still billed for the remainder of that window, so keep the lab dataset small.
If you already run one of these services, you can skip its install step. Our walkthroughs for setting up Backblaze cloud backup and for setting up IDrive cloud backup cover first-time configuration in more depth than this guide can afford.
Prerequisites and Versions
You need a Windows 11 PC you can afford to experiment on, preferably not your daily driver. A spare laptop or a virtual machine works, as long as it has a real internet connection and enough free disk for the lab data. Plan for 20GB free on the C: drive. Everything runs from an elevated PowerShell window, so you also need local administrator rights.
ComponentVersion or requirementWhy it mattersWindowsWindows 11, fully patchedBackblaze and IDrive both support Windows PCsPowerShellPowerShell 7, latest versionFaster hashing with parallel loopsBackblaze client10.0.4.1081Newest confirmed Windows build, fixes install, sign-in and Backup Now problemsIDrive clientLatest version from the IDrive siteRecent release brought a new interface, no numbered version publishedKopiaLatest stable releaseOpen-source client for the Wasabi destinationFree disk space20GB on C:Holds lab data plus a restore targetUpload speedMeasure yours firstSets how long the baseline takes
Record your upload speed before you start. Run a speed test at a quiet time and note the upload figure in Mbps, because the first backup is bound by it. A lab dataset of 8GB at an upload rate of 20Mbps takes around an hour in the best case, and that is before overhead. If your upload is slower, cut the dataset size in Step 1 rather than waiting overnight. The point is a repeatable comparison, not a big number.
Create accounts for all three services before you begin: Backblaze, IDrive and Wasabi. For Wasabi, also create a bucket in an Asia-Pacific region and enable Object Lock at creation time, since you cannot add it later on most S3-compatible platforms. Wasabi’s pay-as-you-go list price is US$7.99/TB/month, so check the price shown for your chosen region before you commit.
Steps 1 to 3: Build the Lab Dataset and Hash Manifest
The lab dataset has to look like a real home folder: lots of small documents, a few mid-sized files, and several large ones. Random bytes do not compress or deduplicate, which makes the upload workload honest. Text files compress well, so a mix shows you how each client behaves with both.
Step 1: Create the lab folder and generate test files
Open PowerShell 7 as administrator and run the generator below. It writes about 8GB by default into C:\BackupLab\Data. Lower the numbers if your upload is slow. The script uses a fixed random seed, so a second run produces the same layout, though the bytes themselves come from the system random generator and will differ.
# New-LabData.ps1 - run in an elevated PowerShell 7 window
$root="C:\BackupLab\Data"
New-Item -ItemType Directory -Path $root -Force | Out-Null
$rng = [System.Security.Cryptography.RandomNumberGenerator]::Create()
function New-RandomFile($path, [int64]$bytes) {
$fs = [System.IO.File]::Create($path)
$buf = New-Object byte[] (1MB)
$left = $bytes
while ($left -gt 0) {
$n = [Math]::Min($left, $buf.Length)
$rng.GetBytes($buf)
$fs.Write($buf, 0, $n)
$left -= $n
}
$fs.Close()
}
# 3,000 small "documents" (4-64 KB) across 30 folders
1..30 | ForEach-Object {
$d = Join-Path $root ("docs\folder{0:D2}" -f $_)
New-Item -ItemType Directory -Path $d -Force | Out-Null
1..100 | ForEach-Object {
$size = Get-Random -Minimum 4KB -Maximum 64KB
New-RandomFile (Join-Path $d ("doc{0:D3}.dat" -f $_)) $size
}
}
# 40 "photos" (3-12 MB)
New-Item -ItemType Directory -Path "$root\photos" -Force | Out-Null
1..40 | ForEach-Object {
New-RandomFile "$root\photos\img$_.raw" (Get-Random -Minimum 3MB -Maximum 12MB)
}
# 6 "videos" of 1 GB each
New-Item -ItemType Directory -Path "$root\video" -Force | Out-Null
1..6 | ForEach-Object { New-RandomFile "$root\video\clip$_.mov" 1GB }
# Text files that compress well
New-Item -ItemType Directory -Path "$root\text" -Force | Out-Null
1..200 | ForEach-Object {
1..2000 | ForEach-Object { "Quarterly report line $_ lorem ipsum backup test" } |
Set-Content "$root\text\notes$_.txt"
}
Write-Host ("Files: {0}" -f (Get-ChildItem $root -Recurse -File).Count)
Expected output ends with a file count a little over 3,200. If the script stalls on the video loop, your disk is probably slow or nearly full. Check free space before you rerun it.
Step 2: Hash every file into a baseline manifest
The manifest is your ground truth. Each row holds a relative path, size and SHA-256 hash. After every restore you will compare the restored folder to this file and count exact matches. Do not skip this step, because file-size checks alone miss silent corruption.
# New-Manifest.ps1
param(
[string]$Root="C:\BackupLab\Data",
[string]$Out="C:\BackupLab\manifest-baseline.csv"
)
Get-ChildItem $Root -Recurse -File | ForEach-Object -Parallel {
$rel = $_.FullName.Substring($using:Root.Length).TrimStart('\')
[pscustomobject]@{
Path = $rel
Size = $_.Length
Sha256 = (Get-FileHash $_.FullName -Algorithm SHA256).Hash
}
} -ThrottleLimit 4 | Sort-Object Path | Export-Csv $Out -NoTypeInformation
Write-Host "Manifest rows: $((Import-Csv $Out).Count)"
Copy the manifest somewhere the attack script cannot reach, such as a USB stick or a second machine. If the simulated attack ever touched it, your comparison would be meaningless. The script in Step 7 only targets the lab data folder, but a separate copy costs nothing.
Step 3: Prepare the timing log
Timing needs to be recorded the same way for every service. A tiny helper that appends a timestamped line to a CSV removes the temptation to eyeball a clock. You will call it at every milestone: backup started, backup finished, attack run, poisoned backup finished, restore started, restore finished.
# Add-LabLog.ps1 - dot-source this: . C:\BackupLab\Add-LabLog.ps1
$global:LabLog = 'C:\BackupLab\timing-log.csv'
if (-not (Test-Path $LabLog)) { 'Utc,Service,Event,Note' | Set-Content $LabLog }
function Add-LabLog($Service, $Event, $Note="") {
$line="{0},{1},{2},{3}" -f (Get-Date).ToUniversalTime().ToString('o'), $Service, $Event, $Note
Add-Content $LabLog $line
Write-Host $line
}
# Usage: Add-LabLog 'Backblaze' 'baseline-start'
Steps 4 to 6: Install the Clients and Take the Clean Baseline
Each service gets one clean baseline of the lab folder. Run them one after another rather than in parallel, because three simultaneous uploads split your bandwidth and ruin the timing comparison. Expect the three baselines to take most of the 90 minutes unless your upload line is fast.
Step 4: Backblaze Computer Backup
Install the Windows client, build 10.0.4.1081 or newer, and sign in. Backblaze’s documentation says the Windows boot drive C: is backed up by default, file sizes are unlimited, and NAS or network-mounted storage is excluded. That default matters here. On a PC full of personal data, the client will start uploading everything on C:, not just your lab folder. Use a fresh VM or spare machine for this test. If you must use a loaded PC, add exclusions in Settings so the client only sees the lab data.
The client also supports files on BitLocker-encrypted drives and lets you add external hard drives, so a lab on a second volume works too. Backblaze describes encryption as 2048-bit public and private keys protecting a symmetric AES-128 key, with an option to add a private encryption key. Leave that option off for this test, since a private key adds a restore step that muddies the timing. Note the choice in your log. Then click Backup Now and call Add-LabLog 'Backblaze' 'baseline-start'. Log the finish when the status shows no remaining files.
Step 5: IDrive Personal
Install the IDrive desktop app and sign in. IDrive recently announced a new user interface, so menu names may differ from older screenshots you find online. In the backup set, choose only C:\BackupLab\Data. Pick the scheduled or continuous option if offered, because the attack test depends on the client noticing changed files. Start the backup manually and log it the same way as Backblaze.
One Australian-specific check belongs here. IDrive’s announcement of new storage regions for cloud-to-cloud backup listed India, Tokyo, Paris, South Korea, South America and South Africa, and that expansion concerns cloud-to-cloud backup rather than the desktop product. Look at the region shown in your account settings and write it down. If data residency in Australia matters to you, do not assume a region. Verify it.
Step 6: Wasabi plus Kopia
Wasabi is storage, not a backup program, so install Kopia and point it at your bucket. Wasabi’s official material describes hot-by-default, S3-compatible storage with no egress or API-request fees, and the platform includes immutability at no extra charge. Create an access key in the Wasabi console, then run the commands below. The endpoint shown follows Wasabi’s regional naming pattern, so confirm the exact endpoint for your chosen Asia-Pacific region in the console. The same tuning thinking applies as in our guide to tuning Backblaze B2 for Australia, since S3 clients behave similarly across providers.
# Create the Kopia repository on Wasabi (Object Lock bucket required for retention flags)
kopia repository create s3 `
--bucket your-lab-bucket `
--endpoint s3.ap-southeast-2.wasabisys.com `
--region ap-southeast-2 `
--access-key YOUR_ACCESS_KEY `
--secret-access-key YOUR_SECRET_KEY `
--retention-mode GOVERNANCE `
--retention-period 720h
# Take the clean baseline snapshot and log it
. C:\BackupLab\Add-LabLog.ps1
Add-LabLog 'Wasabi-Kopia' 'baseline-start'
kopia snapshot create C:\BackupLab\Data
Add-LabLog 'Wasabi-Kopia' 'baseline-end'
# Confirm the snapshot exists
kopia snapshot list
Expected output from kopia snapshot list shows one snapshot for the lab path with a size close to your dataset and a file count matching Step 1. If the create command rejects the retention flags, run kopia repository create s3 --help to see the flags your Kopia version supports. Kopia asks you to set a repository password during creation. Store it in your password manager now, because losing it makes the repository unreadable.
Steps 7 and 8: Run a Safe Ransomware Simulation
The simulation does what real ransomware does to a backup: it replaces files with encrypted copies and deletes the originals, then lets the backup client sync that damage. The script below is deliberately boxed in. It refuses to run against any path except the lab folder, requires an explicit switch, and uses a throwaway key. It is a test tool for your own sandbox, not a malware sample, and you should never point it at real data.
Step 7: Run the attack script on the lab folder
# Invoke-LabAttack.ps1 - encrypts ONLY C:\BackupLab\Data, nothing else
param([switch]$ConfirmLab)
$target="C:\BackupLab\Data"
if (-not $ConfirmLab) { throw 'Pass -ConfirmLab to run the lab attack.' }
if ((Resolve-Path $target).Path -ne 'C:\BackupLab\Data') { throw 'Wrong target.' }
. C:\BackupLab\Add-LabLog.ps1
$aes = [System.Security.Cryptography.Aes]::Create()
$aes.GenerateKey(); $aes.GenerateIV()
[IO.File]::WriteAllBytes('C:\BackupLab\lab-key.bin', $aes.Key + $aes.IV)
Add-LabLog 'ALL' 'attack-start'
Get-ChildItem $target -Recurse -File | ForEach-Object {
$in = [IO.File]::OpenRead($_.FullName)
$out = [IO.File]::Create($_.FullName + '.locked')
$cs = New-Object Security.Cryptography.CryptoStream($out, $aes.CreateEncryptor(), 'Write')
$in.CopyTo($cs); $cs.Close(); $in.Close()
Remove-Item $_.FullName -Force
}
Set-Content "$target\README_LAB_TEST.txt" 'This is a lab simulation. Restore the clean snapshot.'
Add-LabLog 'ALL' 'attack-end'
Write-Host ('Locked files: {0}' -f (Get-ChildItem $target -Recurse -Filter *.locked).Count)
Run it with .\Invoke-LabAttack.ps1 -ConfirmLab. The locked file count should match the Step 1 count. Open File Explorer and confirm that every original is gone and every file now ends in .locked. That is exactly what a victim sees, and it is the state each backup service is about to receive.
Step 8: Let each service back up the poisoned state
This is the step that separates a backup from a recovery plan. Trigger each client and wait for it to finish uploading the encrypted files and recording the deletions. Backblaze and IDrive both have a manual run option, and Kopia needs another snapshot command. Log the finish time for each one. If you skip this step you only test restoring untouched data, which proves little.
. C:\BackupLab\Add-LabLog.ps1
Add-LabLog 'Wasabi-Kopia' 'poison-backup-start'
kopia snapshot create C:\BackupLab\Data
Add-LabLog 'Wasabi-Kopia' 'poison-backup-end'
kopia snapshot list # expect two snapshots now: clean and encrypted
For Backblaze and IDrive, press Backup Now in each app, wait for the idle status, and run Add-LabLog by hand. Now you have the dangerous situation: all three services hold both the clean data and the damaged data, and what matters is how easily you can pull the clean version back out. If a service shows only the .locked files and no older versions, note that immediately as a failed test. That outcome is a finding, not a mistake in your method.
You may like
This is also where immutability earns its keep. A backup that an attacker or a bad script can overwrite is only as safe as its retention settings. We cover the storage-side defence in depth in our guide to immutable backups with Backblaze B2 Object Lock, and the same logic applies to the Wasabi bucket you created with Object Lock enabled.
Steps 9 to 11: Restore the Clean Copy and Verify Every File
Restore each service to its own separate folder so results never mix. Create three empty targets: C:\BackupLab\Restore-Backblaze, C:\BackupLab\Restore-IDrive and C:\BackupLab\Restore-Wasabi. Start a stopwatch the moment you begin each restore and log both ends, because the preparation wait at the service side counts as recovery time in a real emergency.
Step 9: Restore from Backblaze and IDrive
On Backblaze, sign in to the web restore page, browse to the lab folder, and select the file versions from before the attack. Backblaze describes one year of version history and deleted-file retention, but check your account’s version-history setting before the test, since you cannot restore what the setting already dropped. Request a zip restore, which the company says supports up to 500GB per request, then download and extract it into the Backblaze target folder. Log the request time, the ready notice and the extraction finish separately.
On IDrive, open the restore view in the desktop app or web console, choose the earlier version of the lab data, and restore to the IDrive target folder. The new interface may rename options, so look for anything labelled versions, snapshots or restore points. Log the start and finish the same way. If the only choice offered is the latest version, which is the encrypted one, that is a significant failure for this scenario and you should write down exactly what the interface offered.
Step 10: Restore from Wasabi with Kopia
. C:\BackupLab\Add-LabLog.ps1
kopia snapshot list
# Pick the OLDER snapshot ID (the clean one) from the list
Add-LabLog 'Wasabi-Kopia' 'restore-start'
kopia restore SNAPSHOT_ID_CLEAN C:\BackupLab\Restore-Wasabi
Add-LabLog 'Wasabi-Kopia' 'restore-end'
Because Wasabi advertises no egress charges, subject to its fair-use policy, pulling the data back should not add a line to your bill. That is a real advantage over services that meter downloads, though your own download speed still sets the pace. Kopia prints progress as it restores, and a clean run ends without errors.
Step 11: Verify every restored file against the manifest
# Compare-Manifest.ps1
param(
[Parameter(Mandatory)][string]$RestoredRoot,
[string]$Baseline="C:\BackupLab\manifest-baseline.csv"
)
$base = Import-Csv $Baseline
$ok = 0; $bad = 0; $missing = 0
foreach ($row in $base) {
$p = Join-Path $RestoredRoot $row.Path
if (-not (Test-Path $p)) { $missing++; continue }
if ((Get-FileHash $p -Algorithm SHA256).Hash -eq $row.Sha256) { $ok++ } else { $bad++ }
}
$locked = (Get-ChildItem $RestoredRoot -Recurse -Filter *.locked -ErrorAction SilentlyContinue).Count
[pscustomobject]@{
Matched = $ok; Corrupt = $bad; Missing = $missing; LockedLeaked = $locked
PercentIntact = [math]::Round(100 * $ok / $base.Count, 2)
} | Format-List
Run it once per restore folder. A perfect result reads PercentIntact 100 with zero corrupt, zero missing and zero locked files. Anything else tells you where each service falls short. Missing files usually mean the restore selected the wrong version window. Locked files mean the poisoned state leaked into your restore. Corrupt files mean something went wrong in transit or extraction.
This kind of test is the practical core of the Essential Eight backup control. If you want the policy context, our piece on why regular backups is the Essential Eight control ransomware actually tests explains why recovery drills matter more than backup schedules.
For managed fleets, Backblaze has also introduced an advanced installer and command-line interface aimed at IT teams, and its documentation covers silent installs through Microsoft Intune and JumpCloud. That is relevant if you want to repeat this test across several machines rather than one.
Steps 12 and 13: Score the Results and Model Five-Year Cost
Raw numbers need a consistent scoring method or you will end up favouring whichever service felt friendlier on the day. Pull the timing log and the three Compare-Manifest outputs into one scorecard. Weight recovery integrity highest, because a fast restore of the wrong data is worthless. Weight cost last, since all three options land in a similar range for a single PC.
Step 12: Fill in the scorecard
CriterionWeightBackblaze Computer BackupIDrive PersonalWasabi plus KopiaClean version restorable after attack (PercentIntact)40%Your resultYour resultYour resultDeleted files recoverableNot listedYour resultYour resultYour resultBaseline upload time (minutes)10%Your resultYour resultYour resultRestore time, request to extracted (minutes)Not listedYour resultYour resultYour resultEffort and clarity of restore interface10%Your resultYour resultYour resultFive-year cost at your data size10%Your resultYour resultYour result
Score each cell from 0 to 10, multiply by the weight, and add up the column. Keep the notes next to each score. When two services land within a point of each other, the written notes decide, because they capture friction like a confusing menu or a restore that needed support. Our benchmark of cloud backup services uses a similar weighted approach if you want a second reference for how to structure scores.
Step 13: Model five-year cost for your real data size
Your lab used 8GB, but your real library is probably 1TB to 10TB. Prices scale differently across these models. Backblaze Computer Backup charges per computer, so data volume barely matters. IDrive Personal charges for a fixed 5TB allowance. Object storage services charge per terabyte per month. The script below computes five-year totals from the verified list prices and converts at an exchange rate you supply.
# Get-FiveYearCost.ps1
param([double]$UsdToAud = 1.00, [int]$Months = 60)
$wasabi = 7.99 # USD per TB per month, pay-as-you-go
$b2 = 6.95 # USD per TB per month
$bbCB = 9.00 # USD per month, one computer
$idrive = 6.99 # USD per month, 5TB, annual billing, not first-year promo
foreach ($tb in 1, 5, 10) {
[pscustomobject]@{
TB = $tb
Backblaze_CB = [math]::Round($bbCB * $Months * $UsdToAud, 2)
IDrive_5TB = if ($tb -le 5) { [math]::Round($idrive * $Months * $UsdToAud, 2) } else { 'n/a' }
Wasabi = [math]::Round($wasabi * $tb * $Months * $UsdToAud, 2)
B2 = [math]::Round($b2 * $tb * $Months * $UsdToAud, 2)
}
} | Format-Table -AutoSize
With the exchange rate left at 1.00, the output below shows US dollar totals over 60 months at today’s list prices. The B2 figure follows the US$6.95/TB/month rate reported for 2026, which matches Backblaze’s own announcement of a pay-as-you-go storage price increase from US$6. The Wasabi figure uses the US$7.99/TB/month rate that a November 2026 price tracker also lists, and the same number appears in Wasabi’s own pricing notice.
Data sizeBackblaze Computer BackupIDrive Personal (5TB)WasabiBackblaze B21TBNot listedNot listedNot listedNot listed5TBNot listedNot listedNot listedNot listed10TBNot listedNot covered by one 5TB planNot listedNot listed
The table assumes prices stay flat for five years, which they will not. Both Backblaze and Wasabi raised storage prices in 2026, so build in a margin. The Computer Backup column assumes the data sits on one computer or attached external drives, since Backblaze excludes NAS and network-mounted storage. Wasabi’s total also ignores egress, because the company advertises none, but it does enforce a 1TB monthly minimum and a 90-day minimum storage duration, so small datasets cost more per gigabyte than the table suggests.
The pattern is clear. For one computer with several terabytes, flat-rate agent services win on cost by a wide margin. For multiple machines, a NAS, or data you want held under your own encryption keys, object storage with a DIY client costs more but gives you control the agents do not.
Australian Considerations: NBN Upload, Data Location and Privacy
Upload speed is the Australian bottleneck. Most NBN plans are built around download, so your first backup of a large library can take days or weeks. That is why the lab uses 8GB. Once you pick a winner, seed the full library in stages: critical documents first, photos second, video last. Use your measured upload speed to estimate the schedule, divide your data size by your real sustained upload rate, and add 20% for overhead.
Read nextVeeam Fields 2 APJ Execs on Ransomware Recovery [2026]Cybersecurity · Oct 28
Data location needs actual checking. Wasabi prices all regions including APAC at the same rate, so an Asia-Pacific bucket costs no extra, though you should confirm in the console which city hosts it. IDrive’s recent region expansion did not name Australia. Backblaze’s retrieved material did not confirm a Sydney B2 region or endpoint. If your obligations require Australian residency, for instance in health or legal work, confirm the region in writing before you commit. Our guide to HIPAA cloud backup and data residency in Australia walks through how to verify this, and the checks apply to ordinary business data too.
On privacy, remember that agent services handle encryption keys differently. Backblaze offers an optional private encryption key, and losing it means losing access to your data. Kopia encrypts client-side by default, with a repository password you control. For more cloud material on this site, browse the cloud computing section.
Common Pitfalls That Ruin the Test
Running the lab on a loaded C: drive. Backblaze backs up the boot drive by default, so a PC with years of files starts a baseline that takes days and swamps your timing data. Use a fresh VM or exclude everything outside the lab.Restoring only the latest version. After the attack, the latest version is the encrypted one. If you restore it, you will report a perfect restore of garbage. Always pick the pre-attack version explicitly.Running the three baselines in parallel. Simultaneous uploads split your line and make every timing meaningless. Run them one at a time.Forgetting Object Lock at bucket creation. Retention flags in Kopia fail against a bucket created without lock support, and you cannot always retrofit it. Recreate the bucket correctly.Treating promotional prices as renewal prices. IDrive’s first-year offers drop sharply and then revert. Model costs on the standard annual price, as the script does.Restoring over the original folder. Use separate restore targets or you cannot tell which service produced which files, and you may overwrite the manifest comparison.Losing the Kopia repository password. The repository is unreadable without it. Save it before you take the first snapshot.
What Good Results Look Like: Sample Output
Before you trust your own numbers, it helps to know what the tools print. The blocks below show the format of the output, not measured results from any service. Your figures will differ with your line speed, region and the day you run the test. A clean restore prints this from Compare-Manifest.ps1:
Matched : 3246
Corrupt : 0
Missing : 0
LockedLeaked : 0
PercentIntact : 100
A failed restore, where the service handed back the poisoned versions, looks very different. Every original path is missing, and the folder is full of encrypted files:
Matched : 0
Corrupt : 0
Missing : 3246
LockedLeaked : 3246
PercentIntact : 0
A partial result, where some files matched and others are missing, needs a closer look. Check whether the files that changed late in the lab session fell outside the version range you selected. Read the timing log next to the numbers, because it shows what happened between the backup and the restore.
Troubleshooting: 10 Problems and Fixes
Backblaze install or sign-in fails. Update to build 10.0.4.1081 or newer. The November 3, 2026 release fixed installation, sign-in and backup-start problems.Backup Now does nothing. The same release fixed Backup Now and deleted-file handling issues. Update first, then restart the Backblaze service from Services in Windows.IDrive menus may not match tutorials. IDrive recently introduced a new user interface, so older screenshots can look different. Search the app for restore, versions or snapshots instead of following older screenshots.Kopia rejects the retention flags. The bucket probably lacks Object Lock. Recreate it with lock enabled, or drop the retention flags for a non-immutable test.Kopia reports an access denied error. The access key lacks bucket permissions or the endpoint does not match the bucket’s region. Recheck both in the Wasabi console.Kopia says the repository password is wrong. You mistyped it or created the repository with a different one. There is no recovery path, so recreate the repository and re-run the baseline.Attack script throws Wrong target. The lab folder path differs from C:\BackupLab\Data. Edit both the guard and the target, and keep them identical.Hash comparison runs very slowly. Lower the parallel throttle if your disk is a slow USB drive, or raise it on NVMe. The 1GB video files dominate hashing time.Restore folder shows zero files after extraction. The zip extracted into a nested folder. Point Compare-Manifest.ps1 at the inner directory that holds the docs and photos folders.Baseline upload seems stuck. Check whether another service is uploading at the same time, then check the client’s throttle setting. Run one service at a time and pause the others.
If a restore fails in a way none of these explain, capture the client log and the timing CSV before you contact support. Support teams respond faster to a precise timeline than to a general complaint, and the log proves you ran a controlled test.
Advanced Tips for a Sharper Backup Test
Throttle uploads so the PC stays usable
On a home line, a full-speed upload can make video calls stutter. Backblaze and IDrive both include throttle controls in their settings. For Kopia, check kopia repository connect s3 --help for an upload speed cap flag in your version, and set it to roughly 80% of your measured upload rate. That keeps the connection usable while the baseline runs.
Repeat the test with a delayed discovery
Real attacks are rarely noticed within an hour. Run a second round where you wait three days after the poisoned backup before restoring. That exposes retention policies that prune old versions quickly. Any service that cannot hand back the clean version after a few days has a gap you want to know about before an incident, not during one.
Add a second, independent copy
No single online service is a full strategy. Pair an agent-based service with a local drive and keep one copy under immutable storage. The Wasabi plus Kopia setup in this test already gives you a locked copy if you used governance mode, and it can be extended to a compliance-mode retention period for stricter protection. If your work involves regulated data, our Terraform-based test of HIPAA cloud backup services shows how to automate bucket and retention setup so every copy gets identical settings.
Test a single-file restore as a separate drill
Whole-folder recovery and single-file recovery behave differently. Most real incidents are one file, such as a deleted spreadsheet. Time that separately. Delete one document, wait for the next backup cycle, then restore only that file from each service. A service that is quick for bulk recovery but slow for one file may frustrate you more day to day than one with the reverse profile.
Complete Working Project: One Script to Report the Whole Bake-Off
The final piece turns the timing log into a durations report so you do not calculate minutes by hand. Save the script below as C:\BackupLab\Get-LabReport.ps1. It reads timing-log.csv, pairs each start event with its end event per service, and prints the elapsed minutes. Combined with the earlier scripts, you now have a complete working kit: New-LabData.ps1, New-Manifest.ps1, Add-LabLog.ps1, Invoke-LabAttack.ps1, Compare-Manifest.ps1, Get-FiveYearCost.ps1 and this report.
# Get-LabReport.ps1
$log = Import-Csv 'C:\BackupLab\timing-log.csv'
$pairs = @(
@{ Name="Baseline upload"; Start="baseline-start"; End = 'baseline-end' },
@{ Name="Poison backup"; Start="poison-backup-start"; End = 'poison-backup-end' },
@{ Name="Restore"; Start="restore-start"; End = 'restore-end' }
)
$rows = foreach ($svc in ($log.Service | Where-Object { $_ -ne 'ALL' } | Sort-Object -Unique)) {
foreach ($p in $pairs) {
$s = $log | Where-Object { $_.Service -eq $svc -and $_.Event -eq $p.Start } | Select-Object -Last 1
$e = $log | Where-Object { $_.Service -eq $svc -and $_.Event -eq $p.End } | Select-Object -Last 1
if ($s -and $e) {
[pscustomobject]@{
Service = $svc
Phase = $p.Name
Minutes = [math]::Round(([datetime]$e.Utc - [datetime]$s.Utc).TotalMinutes, 1)
}
}
}
}
$rows | Sort-Object Service, Phase | Format-Table -AutoSize
$rows | Export-Csv 'C:\BackupLab\lab-report.csv' -NoTypeInformation
Write-Host 'Saved C:\BackupLab\lab-report.csv'
Log events for Backblaze and IDrive using the same names: baseline-start, baseline-end, poison-backup-start, poison-backup-end, restore-start and restore-end. The report then lines up all three services in one table. Attach the Compare-Manifest output to each row and you have everything the scorecard in Step 12 needs.
When you finish, clean up. Delete the lab folder and restore targets, remove the three test backup sets, cancel any service you will not keep, and delete the Wasabi bucket once its minimum retention window allows. Wasabi’s 90-day minimum storage rule means early deletions are still billed for the remainder, so a small lab dataset is the cheaper choice. Keep lab-report.csv and the manifest in case you want to rerun the test after the next client update.
FAQ: Online Backup Services
What are the best online backup services for a single Windows PC in Australia?
For one computer with a large library, a flat-rate agent service such as Backblaze Computer Backup is usually cheapest, while IDrive Personal suits several devices sharing 5TB. Run the test in this guide to confirm which one restores your own data cleanly.
How long does the first online backup take?
It depends on your upload speed. Divide your data size by your sustained upload rate and add roughly 20% overhead. Many Australian connections upload far slower than they download, so large libraries can take days or weeks.
Is online backup safe against ransomware?
Only if the service keeps old versions long enough and you can restore them. Without version history, a sync-style backup can upload encrypted files over the good copies. The simulation in this tutorial checks exactly that.
Does Backblaze Computer Backup cover external drives and NAS devices?
Backblaze’s documentation says external hard drives can be added to Computer Backup, but NAS and network-mounted storage are excluded. If you need a NAS covered, use IDrive’s business tiers or an S3-style service with a compatible client.
Why use Wasabi with a separate client instead of a ready-made backup app?
Wasabi sells storage, not an application. Pairing it with an open-source client such as Kopia gives you client-side encryption you control and Object Lock for tamper-resistant copies. The trade-off is more setup and a per-terabyte bill.
Will the prices in this guide change?
Yes. Backblaze raised its pay-as-you-go storage prices earlier this year, and B2 is reported at US$6.95 per TB per month, while Wasabi lists pay-as-you-go at US$7.99 per TB per month. Re-run the cost script with fresh prices before you commit.
Can I keep my backup data in Australia?
That depends on the service and region you pick. Confirm the exact storage location in your account and in writing from the provider before you rely on it for regulated or sensitive data.
More on this topicCloud ComputingCybersecuritySoftwareAI & Machine LearningTrending nowTungkillo Fury: 3,300-Member Group Warns Labor [2026]Stockland, AirTrunk Drop 2 Sydney Data Centres [2026]Musk Spent US$474M as Data Centre Backlash Grew [2026]Who Gives A Crap Breach Hits 3 Countries [2026]Best Free Office Suite 2026: Free Office Software Alternatives
Nadia DuboisAI & Innovation EditorNadia Dubois is the AI & Innovation Editor at Tech Insider, where she tracks the rapid evolution of artificial intelligence, from foundation models to real-world enterprise deployment. She previously covered AI and startups for La Tribune and contributed to MIT Technology Review’s European coverage. Nadia specializes in generative AI, AI regulation, and the intersection of technology and European industrial policy. She holds a dual degree in Computational Linguistics and Journalism from Sciences Po Paris.View all articles
——————————————————–
Click Here For The Original Source.
