Three providers · one starting brief

VPS Ubuntu: compare plans and running costs.

Prices checked .
USD, before applicable tax.

For this comparison, the starting brief is at least 1 vCPU, 2 GB-class memory, 40 GB storage, one public IPv4 address and 1 TB monthly transfer. Backups are priced separately below. CPU performance, storage units and backup policies differ: these are suitable options to investigate, not identical machines or a speed ranking.

Selection is based on published specifications and costs; we have not benchmarked these three plans.

On a small screen, scroll the table to compare every column.

2 GB-class Ubuntu VPS shortlist — base cost and backup allowance
Provider / planResources / locationComputeBackup optionBudget with this backupPlan conditions
Vultr Cloud Computevc2-1c-2gb1 vCPU · 2 GB RAM
55 GB disk · 2 TB transferUS regions and Europe
$10.00 / month$2.00 / monthAutomatic backups; 20% of VM price. Keeps the two latest copies.$12.00 / month$144.00 for 12 full months at unchanged rates.Read plan conditions ↓
DigitalOcean BasicRegular · 2 GiB1 vCPU · 2 GiB RAM
50 GiB disk · 2,000 GiB transferUS regions and Europe
$12.00 / month$2.40 / monthWeekly images, four-week retention; 20% of VM price$14.40 / month$172.80 for 12 full months at unchanged rates.Read plan conditions ↓
VDSina Standard1 core · 2 GB1 vCPU · 2 GB RAM
40 GB disk · 32 TB transferAmsterdam 1, Netherlands
$15.00 / 30 days$0.50/day; $15.50 for 31 days$4.08 / 30 daysOne 40 GB retained copy at $0.136/day; weekly creation schedule. See the detailed calculation below.$19.08 / 30 days$19.72 for 31 days; $232.14 for 365 days at unchanged rates.Read plan conditions ↓

What the total includes

Compute, its included disk and transfer allowance, one primary public IPv4 and the named provider backup option. Backup retention differs: two copies at Vultr, four weekly copies at DigitalOcean and a one-copy budget at VDSina. No introductory credit, personal discount or annual commitment discount is assumed. Yearly projections are not prepaid annual offers.

What can change your bill

Applicable tax, paid control panels, server administration, extra IP addresses, transfer overage, additional volumes and independent off-site backup storage are excluded. Stopping a VM is not the same as deleting its billable resources. Verify the selected region, current stock, Ubuntu image, backup retention and final checkout total before ordering.

VDSina's advertised $2.10 entry plan has 1 GB RAM and 10 GB disk, so it does not meet this comparison's brief. Its 2 GB option costs $0.50/day. For US visitors, compare the European location with a US region before choosing on price.

Backup charges: 30-day, 31-day and yearly budgets

We checked the VDSina Standard order form for Amsterdam 1 on . The full backup prices are $0.136/day for 40 GB, $0.272/day for 80 GB and $0.544/day for 160 GB, equivalent to $0.0034 per GB per day for each stored copy. These calculations use full prices, without personal discounts or promotional credit.

Budget assumption: one full-size retained copy, stored throughout the period. The order form offers a weekly backup schedule; the schedule and number of retained copies can be changed in the service settings. Weekly creation does not mean weekly billing: each stored copy has a daily charge. One copy is our calculation assumption, not a confirmed default or a retention guarantee.

On a small screen, scroll to see all budget periods.

Standard plans with one full-disk backup — USD, before tax and optional extras
Plan / diskCompute / 30 daysBackup / 30 daysTotal / 30 daysTotal / 31 daysTotal / 365 days
1 vCPU · 2 GB RAM
40 GB disk
$15.00$4.08$19.08$19.72$232.14
2 vCPU · 4 GB RAM
80 GB disk
$20.10$8.16$28.26$29.20$343.83
4 vCPU · 8 GB RAM
160 GB disk
$40.20$16.32$56.52$58.40$687.66

Calculation: days × (daily compute + disk GB × $0.0034 × retained copies). Round the final period amount to cents. For example, two 40 GB copies cost $8.16 for 30 days; with compute, the budget becomes $23.16. If copies exist for different periods, calculate each copy’s storage days separately.

The 365-day figure is a projection at unchanged daily rates, not 12 × 30 days or an annual prepaid offer. Applicable tax, extra transfer, paid panels, administration, additional disks/IP addresses and independent off-site storage are excluded. A same-datacenter provider snapshot does not replace an independent recovery copy. Confirm the final order total and selected retention settings.

Conditions behind these prices

Open a provider to compare CPU allocation, service guarantees, extra charges and exit conditions for the exact plans above. Checked . Terms marked “not verified” need confirmation before ordering.

Vultr Cloud ComputeRegular · vc2-1c-2gbShared CPU · two backup copies · upgrades only
CPU allocation
Shared CPU: one virtual CPU on the Regular Cloud Compute tier. This is not an Optimized Cloud Compute plan with dedicated CPU resources.
SLA and service credits
The currently published SLA offers 100% network and host-node availability per affected service item. It excludes application/OS/configuration failures, attacks, suspended or halted instances, and specified maintenance/emergency-patch cases; the website, API, DNS and control panel are outside this guarantee. Credits depend on eligible outage duration, from 12 hours to 672 hours of the affected service. Open a support ticket and explicitly request credit: the ticket timestamp marks the outage start for the claim. Credits are not automatic or an application-uptime promise. The page is marked last modified 27 June 2022; checked 9 October 2026.
Transfer and overage
The API lists a 2,048 GB transfer allowance for this plan. Published overage is $0.01/GB above the allocation. The $10 compute price in this comparison applies to the checked US and European regions; other locations can cost more.
IPv6
Available, but not assigned automatically unless enabled during deployment. An existing instance can receive an IPv6 subnet and address through its network settings; check the guest OS configuration as well.
Ubuntu images and architecture
The public OS catalogue lists Ubuntu 22.04, 24.04 and 26.04 LTS as x64 images. This shortlist uses x86_64 / AMD64, not ARM64. Confirm the selected image and plan in your chosen region at deployment.
Backup retention and coverage
The automatic-backup option costs 20% of compute and keeps the two most recent copies. Choose daily, alternate-day, weekly or monthly schedules. Copies stay in the same datacenter on separate storage. They cover the instance filesystem, exclude attached block volumes and are not a guarantee of database consistency.
Resizing and downgrades
Upgrade to a larger plan through Change Plan. In-place downgrades are not supported; moving to a smaller server requires a migration. Mid-cycle upgrades are billed at each configuration’s hourly rate for the time used. Allow a maintenance window and take a recoverable copy first.
Billing while stopped
Stopping or powering off the instance does not stop charges. Destroy the instance to end its compute billing; destruction deletes its data. Check separately retained storage and other billable services before assuming the whole account bill has stopped.
Refunds and trial credits
The published policy says payments for services are final and account credits are non-refundable. No promotional credit is included in our price calculation; any offer has its own eligibility, expiry and usage conditions.

Documents reviewed: Cloud Compute provisioning, resize and IPv6 guides; public Plans and OS APIs; Automatic Backups guide; billing and refund FAQs. Published SLA, retrieved in the browser on 9 October 2026.

DigitalOcean BasicRegular · 1 vCPU / 2 GiBShared CPU · four-week weekly backups · disk growth is permanent
CPU allocation
Basic Droplets use shared CPU. The quoted Regular plan supplies one vCPU; its CPU time can vary with other workloads on the host. Dedicated-CPU Droplet families have different prices.
SLA and service credits
99.99% monthly availability per CPU Droplet. The SLA lists a 100% service credit for affected Droplet resources below that threshold, applied to future invoices, not a cash refund. It covers qualifying infrastructure connectivity or root-disk failures; scheduled maintenance and customer application/configuration faults are excluded. Submit a claim within two billing cycles of the outage month.
Transfer and overage
2,000 GiB included for a full month; outbound overage is $0.01/GiB and inbound transfer is free. Allowance accrues with usage time and is pooled at team level. Public IPv6 traffic also counts, so enabling IPv6 does not remove transfer charges.
IPv6
Available at no extra address charge. Enable it for the Droplet and configure the guest network when needed. Its public traffic remains subject to the transfer allowance and overage rate.
Ubuntu images and architecture
The current Droplet image list includes Ubuntu 22.04, 24.04 and 26.04 LTS with x64 image slugs. The Basic Regular plan here uses x86_64 / AMD64. Application and control-panel compatibility still needs a separate check.
Backup retention and coverage
The quoted $2.40/month option is weekly backups at 20% of compute: four copies retained for four weeks. Backups are crash-consistent disk images in the same datacenter and exclude attached volumes. Daily and usage-based backup products have different pricing and retention.
Resizing and downgrades
Power the Droplet off before resizing. A CPU/RAM-only resize can preserve the existing disk and allow a later smaller compatible plan. Expanding the disk is permanent: it cannot be shrunk, and any future plan must fit that disk. Back up first and plan downtime.
Billing while stopped
Powered-off Droplets remain billable until destroyed. This bundled plan is billed per second, with a minimum of 60 seconds or $0.01, whichever is higher, and a monthly cap. Check separately retained snapshots and volumes before assuming the whole account bill has stopped.
Refunds and trial credits
The published policy does not offer refunds; support may review exceptional circumstances. Promotional credits are separate from a refund promise and are excluded from this comparison’s recurring cost.

Documents reviewed: Choosing the Right CPU Droplet Plan; CPU Droplet SLA; Droplet pricing, images and resize documentation; IPv6 Pricing; Backup Features and Backups Pricing; refund FAQ.

Standard plans1 core / 2 GB · AmsterdamDaily billing · configurable backup copies · several terms need confirmation

Provider: VDSina.

CPU allocation
The plan advertises one 3 GHz core. We did not confirm whether that means shared CPU time or an exclusive physical core. Do not treat the core count as a dedicated-CPU guarantee; obtain the allocation and sustained-use limits before a CPU-heavy deployment.
SLA and service credits
No numeric availability target or service-credit schedule was confirmed in the reviewed Standard plan and .com terms. Request the applicable SLA, exclusions and claim deadline if your project needs a contractual availability commitment.
Transfer and overage
32 TB per calendar month, then $2.25 per additional TB. The advertised 10 Gbit/s port is a connection specification, not a measured speed or dedicated-throughput guarantee.
IPv6
An IPv6 subnet can be ordered in the server’s IP tab and then configured in the operating system. Its current price and included allocation were not verified in the .com documentation; confirm both before treating IPv6 as included.
Ubuntu images and architecture
Standard pricing lists Ubuntu 20.04, 22.04, 24.04 and 26.04 and permits a custom ISO. The equipment FAQ identifies AMD EPYC for Standard servers, indicating x86_64 / AMD64 hardware; confirm the exact image architecture at checkout. For 20.04, check extended security coverage before choosing it.
Backup retention and coverage
The checked full price is $0.136/day per retained 40 GB copy ($4.08 for 30 days), equivalent to $0.0034/GB/day. The comparison budgets one full-size stored copy; this is not a confirmed default retention setting. Paid snapshots and scheduled copies are available. The schedule documentation shows daily, weekly and monthly options with a configurable number of copies to keep; no fixed retention period or current maximum was verified. A restored server must be in the same datacenter and have at least the snapshot’s disk allocation. Confirm separate-disk coverage and the selected number of retained copies.
Resizing and downgrades
Resources can be increased; disk growth may require expanding the guest partition/filesystem and a restart. The provider’s downgrade Q&A says Standard plans cannot be reduced in place, unlike its separate Cloud product. That answer is older documentation: confirm current options before choosing a plan you may need to shrink.
Billing while stopped
Daily service renewal is separate from an OS shutdown. The Daily Use FAQ describes disabling Auto-prolong: service stops after a day and is deleted if not reactivated within five days; this option is limited to once per calendar month. It is not indefinite free storage for a stopped VM.
Refunds and trial credits
The Standard FAQ offers no test period. The .com terms say cancellation takes effect at the end of the current term and subscription fees are not refunded, subject to applicable law. A daily price is not a money-back guarantee; do not assume unused prepaid funds will be returned.

Documents reviewed: Standard server pricing and image list; order-form backup quote for Amsterdam 1, checked 9 October 2026; equipment, IPv6, scheduled backup, snapshot, disk expansion, downgrade and Daily Use Q&A; provider Terms sections 5–6. Some Q&A instructions predate this review.

Before you order: record the final checkout total, the number of recoverable copies and the steps that end billing. Keep an independent recovery copy when a datacenter-wide failure would be unacceptable.

View server plans ↗

Choose resources for the application

Ubuntu VPS configurations:
where to start.

Planning estimates for an early-stage workload.
Not provider plans or measured performance.

Choose the software you need to run before choosing a server size. A database, a control panel and background workers share the same resources as your application. The estimates below match the starting budgets in our resource selector; they do not predict how many visitors a server can handle.

Initial resource budgets — validate with your own workload
ProjectvCPURAMSSD / NVMeWhat to measure first
Linux learning lab11–2 GB20 GBMemory during package installation; available space. Heavy builds or a desktop need a separate budget.
Small dynamic website22–4 GB30 GBCMS and database memory, uncached requests and the size of uploaded files.
Web application or API24 GB40 GBConcurrent requests, worker memory and queue growth during busy periods.
Shop with a database48 GB80 GBCheckout latency, database queries, catalogue imports and scheduled jobs.

These are examples for planning, not minimum requirements. A static site may need no VPS at all, while an application with image processing or a large database can exceed these budgets. Leave room for updates, logs and data growth; keep independent backup storage in your cost estimate.

Who does the maintenance?

Choose the support scope as carefully as the server.

The prices above are infrastructure budgets. Do not assume that a server control panel or 24/7 technical support includes application administration. Check the service agreement before relying on someone else to patch, monitor or recover your stack.

Responsibilities to confirm before ordering
TaskSelf-managed VPSWhat to ask about a managed service
Physical host and provider networkProvider operates its infrastructure under the service terms.Check availability commitments, maintenance notices and escalation.
Ubuntu, firewall, application and databaseYou arrange configuration, updates and monitoring.Which packages, applications and incident types are included?
Backups and recoveryYou choose the copies, verify them and plan recovery.Who performs restores, what data is covered, how long are copies kept and what costs extra?

If you cannot maintain an operating system, include a management service in the budget. For a site with only static files, compare static hosting first. For free experiments, check allowances, expiry and eligibility.

For visitors in the United States

US or Europe: choose an Ubuntu VPS region.

Location information checked
.

For a shop, API or logged-in app serving mainly US users, start with a US region near your largest user group. A European origin can suit a mostly static site with a CDN, or an application whose users and database are mainly in Europe. These are starting points for selection; we have not run the same application on matched US and European servers.

Choose a city as well as a continent. An East Coast region is a candidate for an East Coast audience; a West Coast region is a candidate for a West Coast audience. For a nationwide audience, check both coasts and a central US location before choosing.

On a small screen, scroll the table to compare the options.

Region selection by workload — guidance, without a measured speed ranking
Your workloadA US origin is a candidate when…A European origin is a candidate when…Check before choosing
Shop, API or account dashboardMost interactive users are in the US and the database can run near the application.Most interactive users and the main database are in Europe.Login, search, cart and checkout responses that reach the origin; include errors and slow responses.
Static site, documentation or image catalogueUS visitors frequently request uncached pages or freshly published content.A CDN can serve the cacheable content near visitors and origin requests meet your response-time target.Both cache hits and misses, including the first request after an update.
Mixed US and European audienceUS users account for more of the transactions that need a server response.European users account for more of those transactions.Results for each audience separately; a combined average can hide a slow region.
Application with an existing databaseThe primary database and critical services are already in the US.They are already in Europe.The complete application-to-database path. Moving only the web server can add cross-region database calls.
Scheduled jobs and background processingThe source data and destination services are in the US.The source data and destination services are in Europe.Job completion time, data-transfer charges and resource use; visitor proximity may be secondary.

What the checked providers offer

Vultr lists US locations including New Jersey and Los Angeles, plus Amsterdam in Europe. DigitalOcean lists New York and San Francisco in the US, plus Amsterdam (AMS3) in Europe. The VDSina Standard offer compared here is Amsterdam 1, Netherlands. A US CDN node does not change that origin location. Confirm the exact plan and current capacity in the selected city.

Location sources: Vultr region catalogue, DigitalOcean Regional Availability and the published Amsterdam location information. Provider locations are facts from those sources; the workload choices above are our editorial guidance.

Before moving an application: a practical checklist
  1. Identify the cities producing your important transactions and where the database and external services run.
  2. Set a response-time target and acceptable error rate for login, search and checkout. Check those actions with caching bypassed where appropriate.
  3. For a direct comparison, keep the application, data, Ubuntu release, resources and load the same. Compare from the same client locations, separating cache hits, misses and direct-origin requests.
  4. Record p50 and p95 latency, sample counts, errors, CPU and RAM over repeated runs. These are measurements to collect, not results we have obtained for a US/EU server pair.
  5. Choose against your targets and full budget, including backups and transfer. Before switching traffic, check data consistency and keep a rollback path.

Existing measurements: delivery of this website

The observations below cover one website delivered through Cloudflare. There was no US application server in this test, so the numbers cannot identify a faster US or European application deployment.

Our US delivery test: 36 requests, 9 October 2026

We requested this site's homepage and its 13,358-byte AVIF through Globalping from two probes in each city, in three rounds. Every response was HTTP 200. The homepage returned CF-Cache-Status: DYNAMIC; that describes cache handling, not a database-backed application. The asset returned HIT in all 12 requests from rounds two and three.

Response wait in milliseconds: median (observed range)
Probe cityHomepage, 6 requests per cityWarm asset, 4 requests per city
New York523.5 ms (512–706)15 ms (11–22)
Dallas665.5 ms (657–956)25 ms (23–29)
Los Angeles778 ms (655–994)19.5 ms (17–28)

The metric is Globalping's firstByte wait phase; DNS, connection and TLS timings are separate in the data. Probes and routes can differ between rounds. This small sample is not browser LCP, uptime, server capacity or a comparison of the three providers. It does not predict a database application's response time.

Download observations and measurement IDs (JSON) · View the first homepage measurement. Recheck your own application from the same regions before committing.

Read the hosting evaluation method for resource, cost and network checks.

Release, application, provider

Choose an Ubuntu image your stack supports.

Documentation checked
.

Start with the application and its extensions, then choose a supported Ubuntu release and CPU architecture. The matrix records published support; it does not claim that we installed every combination.

Scroll sideways on a small screen to compare all columns.

Documented Ubuntu LTS compatibility
ReleaseStandard maintenanceVDSina standard imageDocker EngineHestiaCP fresh install
26.04 LTSUntil 2031ListedSupportedSupported
24.04 LTSUntil 2029ListedSupportedSupported
22.04 LTSUntil 2027ListedSupportedSupported

Sources: Canonical release lifecycle · Provider images and application templates · Docker OS requirements · HestiaCP requirements.

A supported panel is not a ready-made template

Hestia documents all three releases above, but the listed Hestia application template is labelled Ubuntu 22. A plain 24.04 or 26.04 image and a supported manual installation are a different path. Confirm the exact template at checkout.

Architecture and resources still matter

Hestia requires a fresh 64-bit AMD64 or ARM64 system and recommends 4 GB RAM. Check every container image, native extension and recovery agent for your chosen architecture. Docker support does not establish compatibility of the application inside it.

Plan how the server will be maintained

Standard Ubuntu security maintenance covers Main; Ubuntu Pro and third-party software have separate coverage. Test release upgrades on a copy, including a database operation and restore, before moving live traffic.

Use the image and compatibility guide to identify the delivered image, inspect package versions and record an upgrade decision.

Start with the workload

Match resources
to a checked plan.

Choose a project type for an initial estimate.
Validate it with your application under load.

The larger example doubles CPU, disk and the upper RAM estimate. It is a budget comparison, not a traffic forecast or hosting plan.

Your initial resource budget
2vCPU
2–4GB RAM
30GB SSD / NVMe

For a small dynamic site. Measure the memory used by your CMS and database.

View VPS plans

See matching configurations ↓

Choose the configuration and region on the provider website; these selections are not transferred.

Published plans that meet this resource budget

We match the upper RAM estimate, CPU count and disk allocation to the checked plans below. Matching specifications does not establish application performance. The shortlist covers three sizes per provider, not their entire catalogues.

For the default 2 vCPU / 4 GB / 30 GB budget, these three offers meet the listed specifications.

  • Vultr Cloud Compute

    2 vCPU · 4 GB RAM · 80 GB disk

    $20.00 / month compute

    $24.00 / month with automatic backups.US regions available; select one at checkout.Read plan conditions ↓
  • DigitalOcean Basic Regular

    2 vCPU · 4 GiB RAM · 80 GiB disk

    $24.00 / month compute

    $28.80 / month with weekly image backups.US regions available; select one at checkout.Read plan conditions ↓
  • VDSina Standard

    2 vCPU · 4 GB RAM · 80 GB disk

    $20.10 / 30 days compute

    $28.26 / 30 days with one retained 80 GB backup copy.Amsterdam 1, Netherlands; not a US server.View server plans ↗

Prices checked 9 October 2026. Before tax; no paid panel, management or independent off-site copy included. The Amsterdam Standard plans use daily billing. See the comparison assumptions and backup sources.

Published application options

Provider: VDSina. Product information checked .

Provider descriptions for choosing a server family
Server familyPublished specifications
StandardPositioned for general-purpose tasks, with a 3 GHz processor and NVMe storage. The three Standard configurations in our shortlist have 1 / 2 / 4 cores, 2 / 4 / 8 GB RAM and 40 / 80 / 160 GB disks respectively. Each lists 32 TB monthly transfer and a 10 Gbit/s internet port.
Hi-CPUPositioned for tasks that demand CPU speed, with a processor frequency up to 4 GHz. This is a separate server family from the Standard offers in our comparison.

Checking your server’s load: The provider’s help guide directs customers to My servers → select the service → Statistics in its control panel to review load indicators.

Historical provider performance results — 2019

In its company article dated , the provider reported a Bitrix performance score of 107 for its then-Standard setup on Intel Xeon Silver 4116. For the Hi-CPU setup described there, it reported at least 30% more Bitrix performance under high server load and around 60% in favourable conditions.

The article also includes UnixBench 5.1.3 logs with a Linux 3.10 el7 kernel. These historical provider results describe that hardware and configuration; they do not establish the capacity of today’s Ubuntu Standard plans.

Source: the provider’s company blog, “How to launch a Hi-CPU VDS for Bitrix”, 12 September 2019.

Sources: Standard and Hi-CPU product pages; “How do I see the server load report?” help guide. These are provider-published specifications and results. Our resource selector remains a planning estimate; we have not measured application capacity on these plans.

Evidence from our own site

Recovery checks: Ubuntu application and site files.

Ubuntu application and PostgreSQL: a completed recovery drill.

On , we restored a small notes application, its PostgreSQL database and three uploaded files into a separate clean Ubuntu VM. The source VM was powered off before recovery. This was a synthetic application built for the drill; this website itself serves static files.

Recreated the environment

Ubuntu 24.04.5 LTS, Nginx 1.24.0, Python 3.12.3 and PostgreSQL 16.15. Both VMs used 2 emulated vCPUs, 1,536 MiB RAM and an 8 GiB virtual disk. OS, kernel and installed package versions matched; machine identities differed.

Recovered the data

All 1,001 database records matched the original ordered-data checksum. Application files and three uploads matched their checksums and permissions. The database sequence was restored; the application secret was supplied separately.

Verified a new transaction

HTTP checks read the recovered records and uploads. An unauthorized write was refused. A new record was created through the application, then remained readable after a full Ubuntu reboot and automatic service startup.

Recovery steps, observed times and limits
  1. Pause application writes, create a PostgreSQL custom-format dump and archive the app, uploads, Nginx and systemd configuration. Record package versions and checksums.
  2. Copy the bundle out of the source VM and verify its SHA256. The integrity check rejected an intentionally damaged copy.
  3. Power off the source. On the clean Ubuntu target, install the recorded package versions and supply the secret separately with root-only file permissions.
  4. Verify the bundle, recreate the application user and database, restore the dump in one transaction, correct file ownership and start the services.
  5. Compare all restored rows and files, test HTTP reads and an authenticated write, then reboot and check that the new record persists.

The backup has a cutoff. A record written on the source after backup was correctly absent from the restored database. The new post-recovery transaction reused the next available ID. Backup frequency and a verified recovery procedure determine which later writes you can recover.

Observed on this local run: 42.7 seconds for the restore script; 481.0 seconds from the already booted clean target through environment installation, bundle transfer, restore and first checks. These figures exclude image download and initial boot. They are not a promised recovery time.

Scope: QEMU software emulation on Windows, with Ubuntu running inside each VM. This demonstrates recovery of this example environment, application and logical database backup. It does not test a provider backup-panel restore, a whole-disk snapshot, DNS/TLS cutover, external integrations or a production outage.

Read the recovery evidence and HTTP results (JSON).

Static files of this website: a separate restore check

This site serves prebuilt HTML through Nginx and Cloudflare. Its page requests do not use an application database. Our evidence describes that workload; it cannot establish the capacity of a WordPress site or shop.

Current first-party files

Shared CSS: 21,407 bytes. Shared JavaScript: 3,141 bytes. Homepage additions: 5,228 bytes CSS and 6,086 bytes JavaScript, before transfer compression.

Every restored file checked

On 9 October 2026, we restored the archived 1 October release to an empty local test directory. All 1,085 files matched their SHA256 manifest. An intentional change to a test copy was detected, then corrected.

Keep the scope clear

The restore used a Windows filesystem and a Python HTTP server on loopback. Six HTTP responses also matched the expected files. This did not recover a VPS disk, start Nginx or restore a database.

Inspect the restore method and its limits

The archive contained 6,665,440 bytes; the restored release contained 34,282,070 bytes. Extraction took 2.104 seconds and the first complete hash check took 36.2 seconds on this local run. These are observed file-operation timings, not an outage recovery commitment.

  1. Start from an existing release archive and its separately recorded file manifest.
  2. Restore only regular, expected files into a new empty directory; reject paths outside it.
  3. Compare every restored file with the expected SHA256.
  4. Check that a deliberately altered test file fails verification.
  5. Serve the restored copy locally and compare representative HTML, CSS, JavaScript and image responses.

Read the verification record (JSON). The active site was not changed by this exercise. HTML, analytics and third-party requests are excluded from the asset figures above.

For your own database application, repeat the check with its database export, uploads, secrets supplied separately, permissions and a real transaction after recovery. Define acceptable data loss and recovery time before choosing a backup schedule. Use the backup and restore guide.

Before ordering

Ubuntu VPS questions.

More questions →
How much does an Ubuntu VPS cost with backups?

In our checked 2 GB-class shortlist, Vultr is $12/month with its automatic-backup option and DigitalOcean is $14.40/month with weekly backups, before applicable tax and optional extras. VDSina is $19.08 per 30 days: $15.00 compute plus $4.08 for one retained 40 GB backup copy. Its 31-day total is $19.72. Backup retention differs between these offers. See the dated assumptions and sources.

Is 2 GB of RAM enough for my application?

It depends on the application, database, workers and peak load. The estimator is a planning budget, not a visitor-capacity test. Match the full stack and measure memory, latency and errors before choosing a smaller plan.

Can I choose a server in the United States?

US regions are available in the checked Vultr and DigitalOcean catalogues. The listed VDSina offer is in Amsterdam. Choose US region required in the plan selector and confirm the region at checkout.

Does the provider maintain Ubuntu and my website?

Read the support scope. An unmanaged server generally leaves OS and application administration to you. A control panel is a management tool; it does not by itself mean that an administrator maintains the server for you.

Which Ubuntu release should I choose for Docker or a panel?

Use the documented compatibility matrix, then check your own application and architecture. A provider’s preinstalled panel template can use an older release than the panel supports for a fresh manual installation.

Can I start with a free Ubuntu VPS?

Free allowances and trial credits have different time limits, quotas and billing conditions. Read the free-options guide. A local virtual machine can also suit Linux practice.

Who operates this website and handles my server order?

VPS Ubuntu is an independent publisher operated by Metro Manhattan Office Space, Inc., with articles under the name VPSuntu. The selected hosting provider handles orders, payments and server support.

Make a documented choice

Match the plan to your project.

Compare resources, region, backup terms and the complete bill before ordering.

Explore servers ↗

Review the comparison ↑