Hosting decisions look technical until unstable infrastructure starts shaping lead cost, speed, uptime, and release risk.
For a commercial website, the line between shared hosting and VPS is often the line between a system that can absorb growth and one that fails when growth arrives.
This article explains when VPS becomes the rational baseline, what shared hosting still does well, and how to choose based on business load rather than habit.
Why the question âVPS or sharedâ is more important than it seems
Hosting decisions look technical until unstable infrastructure starts shaping lead cost, speed, uptime, and release risk.
For a commercial website, the line between shared hosting and VPS is often the line between a system that can absorb growth and one that fails when growth arrives.
This article explains when VPS becomes the rational baseline, what shared hosting still does well, and how to choose based on business load rather than habit.
The main thesis of 2026
A separate cloud VPS for each commercial website is not a âprivilege of large projectsâ, butbasic infrastructure quality standard. Shared hosting in 2026 only makes sense for personal blogs, landing pages without traffic and temporary demos. As soon as a site receives an advertising budget or SEO goals, saving 5-10 BYN per month on shared turns into hundreds and thousands of BYN per month in lost revenue due to ânoisy neighborsâ, slow TTFB and stack limitations.
How are shared and VPS physically different?
To understand where the difference in speed and stability comes from, itâs useful to take one look at the two types of hosting âfrom the inside.â Both run on the same hardware - a physical server in a data center. What is fundamentally different is how this hardware is divided between projects.
Shared hosting- this is one physical server on which several hundred (sometimes several thousand) websites of different clients simultaneously live. They all share one CPU, one RAM, one disk array. The isolation between them is minimal: containers or chroot, but not full virtualization. If a neighbor on the server launched a heavy import, fell under DDoS, or his WordPress with 50 plugins is glitchy, your site slows down along with it, even if your code is perfect and there is almost no traffic.
VPS(Virtual Private Server) is a dedicated virtual machine with guaranteed resources (CPU, RAM, disk, network), running on a physical server via a hypervisor (KVM, VMware, HyperâV). Dedicated means truly yours: no other client can take your processor or memory. You have your own full Linux (or Windows), your own settings for nginx, PHP, database, caching, firewall. The load of neighbors on the physical server does not affect you (with the exception of rare ânoisy hypervisorâ problems, which serious providers encounter once a year).
In practice, this means: on shared hosting, you do not control a single component of your stack. PHP versions, memory limits, caching policy, nginx rules, access to logs - all this is determined by the provider, and usually conservatively and with a reserve âfor the worst neighborâ. On a VPS, all these decisions are made by you (or your contractor): you can enable HTTP/3, tweak OPcache, install Redis for the cache, configure individual nginx rules for your CMS, monitor logs in real time.[6]
Four stages of website growth and when the type of hosting changes
The journey of a commercial website from start-up to maturity can be described in four stages based on the type of infrastructure required. This is not a rigid scale, but an empirical guide: what stage your project is at and what type of hosting it needs.
Four stages of growth and the corresponding type of hosting
Stage 1
Promo site / landing page without advertising
Business card website, promo page, test landing page. There is practically no traffic, income does not depend on the site. Shared hosting or even free hosting is suitable without reservations. A high-quality VPS here is redundant and more expensive.
Stage 2
Start of a commercial project with advertising
The site is launched with an advertising budget, the goal is to receive applications. Even 1,000â5,000 sessions per month already pose challenges for speed, Core Web Vitals, and stability. From this stagea separate cloud VPS becomes the basic minimum: 15â30 BYN/month against hundreds of lost BYN revenue.
Stage 3
Mature project with SEO and traffic
10â50 thousand sessions per month, active SEO, developed catalog, integration with CRM/ERP. VPS is a must, with a reserve of resources: 4â8 GB RAM, SSD NVMe, proper caching, CDN for static data, regular backups, uptime monitoring.[2]
Stage 4
Large project / high loads
Hundreds of thousands of sessions per month, tens of thousands of pages, peak loads from advertising. One VPS is no longer enough: separate servers are needed for the database, application, cache; balancer, scaling; perhaps Kubernetes containers or managed solutions.
Critical to remember:The transition from stage 1 to stage 2 is not a gradual âweâll grow up and move,â but a conscious transition even before the first advertising ruble. A typical mistake is to launch a site on shared âas a testâ, pour in advertising traffic and discover that with 100â200 simultaneous visitors, shared begins to choke, the form freezes, and Core Web Vitals fall apart. By this point, the advertising budget has already been eaten up, applications have been lost, and the move to VPS is in emergency mode. It makes more sense to start immediately from the second stage, even if the site is new.
Eight technical advantages of VPS that are not shared
The difference between shared and VPS is not âone point in the configurationâ, but a set of key features, each of which directly affects the speed, stability and security of a commercial website. Below is a minimum set of advantages that are either impossible or severely limited on shared hosting.
Dedicated CPU and RAM
Guaranteed resources that cannot be âeatenâ by a neighbor on a physical server. This means predictable TTFB and stable LCP under any load within your tariff.[2]
Full control over the stack
Your choice: Ubuntu/Debian, nginx or Apache, PHP 8.3 or Node.js, MySQL or PostgreSQL, Redis for cache, any modules and extensions. On shared versions and settings are fixed by the provider and rarely change.
Multi-layer caching
OPcache for PHP, object cache (Redis/Memcached), nginx HTTP cache (fastcgi cache), CDN for static. On shared, a maximum of 1â2 levels are usually available, and without fine tuning the cache works worse than it can.[6]
Normal access to logs
Access log, error log, slow query log, PHP errors - all in real time, you can analyze tail, grep, Kibana, by any means. On shared, access to logs is often limited or provided with a delay.
Firewall and DDoS protection
Own iptables/nftables, Fail2ban, integration with Cloudflare or other CDNs for DDoS protection, rate limiting at the nginx level. On shared, you depend on the general policy of the provider and often cannot âspecificallyâ prohibit suspicious traffic.
Regular automatic backups
Snapshot of the entire virtual machine from the provider + your own scripted backups of the database and files to separate storage (S3/Backblaze/separate VPS). Recovery is done in minutes, not in hours of correspondence with shared support.
HTTPS and modern protocols
Own Let's Encrypt certificates with auto-renewal, HTTP/2 and HTTP/3, OCSP stapling, modern TLS configuration, free choice of cipher suite. On shared, the protocol stack is limited and often lags 2-3 years behind current standards.
Monitoring and alerts
Uptime monitoring through external services (UptimeRobot, Pingdom, StatusCake), internal monitoring (Netdata, Prometheus, Grafana) with alerts in Telegram or email. You learn about the problem in seconds, and not from user complaints.
Comparison of shared and VPS by key parameters
Below is a summary table that helps you quickly see the difference between the two types of hosting in terms of those parameters that really affect a commercial website. This is not an exhaustive technical reference, but a practical guide for decision making.
| Parameter | Shared hosting | Cloud VPS |
|---|---|---|
| Resources | shared with hundreds of neighbors | dedicated: CPU, RAM, disk, network |
| Typical TTFB | 800â2500 ms under load | 150â400 ms stable[2] |
| Control over the stack | no or minimal | full: Linux, nginx, PHP, DB |
| Caching | Level 1, often without fine tuning | OPcache + Redis + fastcgi cache + CDN[6] |
| Access to logs | limited, through panel | full, real-time, SSH |
| HTTPS and protocols | TLS 1.2/1.3, HTTP/2 - if the provider has enabled | TLS 1.3, HTTP/2, HTTP/3, OCSP - your choice |
| DDoS protection | general provider policy | your own rules + CDN + fail2ban + rate limit |
| Backups | according to the provider's schedule, often weekly | snapshot + your own script, up to hourly |
| Scalability | just âmove to another tariffâ | vertically in minutes, horizontally in hours |
| Core Web Vitals | itâs difficult to get into the green zone | the green zone is achievable with the standard setting[1] |
| Price in 2026 | 5â15 BYN/month | 15â120 BYN/month depending on resources |
| Suitable for | personal blogs, landing pages without traffic | any commercial website with advertising and SEO |
It is very important to pay attention to two points from the table:TTFBAndCore Web Vitals. These are not âtasteâ parameters, but metrics that Google directly uses in ranking through Page Experience and in calculating Landing Page Experience in Ads.[3]The difference of 1â2 seconds TTFB between shared and VPS is the difference between âa site in the top search results for competitive queriesâ and âa site in the second ten,â even with the same quality of content.
Six myths about VPS that are why people still choose shared
Despite the obvious difference, some businesses and developers still prefer shared. Usually, it is based on several persistent myths that were true 5-10 years ago and no longer correspond to the reality of 2026.
Myth: âVPS is expensiveâ
not anymore
Myth: âVPS requires a separate adminâ
partially true
Myth: âShared is easierâ
no longer a fact
Myth: âShared is also fastâ
just until the neighbors show up
Myth: âCloudflare will solve all problemsâ
partially
Myth: âShared is safer, everything is set upâ
more myth than truth
TCO formula - comparison of real costs over a three-year horizon
The most common argument in favor of shared is âcheaperâ. But if you count not the monthly bill, butTotal Cost of Ownership(total cost of ownership) for 3 years, taking into account lost revenue, the picture changes to the opposite.
A typical scenario for an average commercial website with advertising.Shared: 10 BYN ÃÂ 36 = 360 BYN for hosting + 0 BYN setup + ~50,000 BYN loss of revenue from slow work for 3 years (conservatively, with a conversion of 1.2% instead of a possible 2.0% on a fast site) + ~3,000 BYN emergency move to VPS after 1.5 years, when problems become obvious =~53,360 BYN TCO. VPS: 60 BYN ÃÂ 36 = 2,160 BYN for hosting + 800 BYN one-time setup + 0 BYN losses =2,960 BYN TCO. The difference is approximately 18 times in favor of VPS.
To this calculation it is worth adding three factors that are poorly digitized, but still work against shared. The first is the cost of a click in advertising: a poor Landing Page Experience due to a slow server can increase CPC by 30â70%, which over the course of a year of the campaign results in tens of thousands of BYN in overspending.[4]The second is SEO positions: a slow site grows slower in organic rankings, and this is lost traffic that cannot be returned. Third, reputational damage from downtime: every âneighbor fell, so did Iâ episode, it is your clients who lose, not the clients of the shared provider.
Four scenarios: when shared is enough, and when VPS is required
The question of choosing a hosting is not an absolute âVPS is always betterâ, but a selection for the scenario. Below are four typical situations and recommendations.
Personal website / promo page
shared is enough
New commercial website with advertising
VPS only
Mature site with traffic
VPS is required
Highly loaded project
VPS + scaling
outdated approach
âWeâll launch it on shared, if it goes well, weâll move to VPSâ
The site starts on shared âto save money.â Advertising does not pay off because TTFB 1.5â2.5 s breaks Core Web Vitals and Quality Score. SEO is not growing for the same reason. Six months later, the business decides âeverything is bad because of the site / advertisingâ - and either closes the project or makes an emergency move to a VPS, losing 10â50 times more in 6 months than the VPS would have cost from the very beginning.
modern standard
âA separate cloud VPS for each commercial project - from day oneâ
The site starts on a VPS with dedicated resources, the right stack (nginx, PHPâFPM, OPcache, Redis, SSL), CDN for static content, regular backups and monitoring. TTFB 150â400 ms, Core Web Vitals in the green zone, Landing Page Experience high. Advertising pays off, SEO grows, business scales - without emergency relocations.[5]
Why VPS has become so much more affordable
Over the past 5 years, there have been three changes that fundamentally changed the economics of VPS. First, cloud providers have divided servers into rental minutes: a starting VPS now costs 4â8 USD/month, and not 50â100 USD, as before. Second, they appearedready-made images(DigitalOcean Marketplace, Hetzner Apps) with a pre-configured stack: one-click deployment. Third, standardization of management (Ansible, Docker, ready-made scripts) reduced setup from days to hours. Result:a separate VPS for each client site is the norm today, not exotic, and this is how the infrastructure of all commercial projects should be built.
Checklist: ten points for correctly ordering and setting up a VPS
If you decide in favor of VPS, it is important to immediately lay down a minimum of practices, without which VPS will simply turn into âmore expensive sharedâ. Below are 10 points that you should go through when launching a new VPS for a commercial website.
Correct VPS start - 10 points
- Selecting a provider with the desired location:the closest data center to the audience (for Belarusian traffic - Minsk/Warsaw/Frankfurt), with an SLA of 99.9%+ and transparent scaling rules.
- Configuration with reserve:at least 2 CPU / 4 GB RAM / 40 GB NVMe SSD for a production site on a CMS; for large ones - 4 CPU / 8 GB / 80 GB SSD and higher.
- Current OS:Ubuntu LTS or Debian stable; no âlegacyâ distributions that are about to run out of security support.
- Modern web server:nginx with HTTP/2 and HTTP/3, proper settings for keepalive, gzip/brotli, fastcgi cache.[6]
- Current PHPâFPM (or Node.js):PHP 8.2/8.3, OPcache enabled, JIT configured, realpath cache increased to fit the project size.
- Multi-layer caching:OPcache for code, object cache (Redis/Memcached) for application, nginx HTTP cache for pages, CDN for static content.[5]
- Let's Encrypt with auto-update:HTTPS by default, auto-renewal of certificates via certbot, OCSP stapling, HSTS - everything is modern.
- Basic firewall and security:ufw/nftables, key SSH authorization without password, separate SSH port, fail2ban, automatic security updates.
- Regular backups:daily snapshot from the provider + your own scripted backups of the database and files to separate storage (S3 / external VPS / Backblaze B2).
- Uptime and resource monitoring:external uptime monitoring (UptimeRobot / StatusCake), internal (Netdata / Prometheus + Grafana), alerts in Telegram / email for any anomaly.
How Ontop solves this
In our team, a separate cloud VPS for each clientâs commercial site is not a âpremium optionâ, but a standard basic configuration. We donât launch sites on shared hosting because weâve already seen how it ends many times: slow TTFB, problems with Core Web Vitals, expensive advertising, inability to scale, emergency moves. It's cheaper and easier to do it right once.
Technically, our standard stack is Ubuntu LTS, nginx with HTTP/3, PHPâFPM 8.3 with OPcache and JIT, MySQL 8 or PostgreSQL, Redis for object cache, fastcgi cache nginx for HTML cache, Cloudflare as a CDN, Let's Encrypt for HTTPS, automatic backups to separate storage, monitoring through external and internal services. This configuration provides a stable TTFB of 150â300 ms, green Core Web Vitals and can withstand peak loads from advertising campaigns without degradation. This is the very basis from whichthe payback of speed and hosting in applications and conversion depends.
A separate VPS for each client gives us another important advantage - a clear division of areas of responsibility and security. Problems of one site do not affect others, backups are independent, settings are individual for a specific CMS and load. If necessary, we can carry outSEO migrationsorcomplete redesignwithout risk to other projects. And when a site outgrows one machine, vertical scaling (more CPU/RAM) is done in minutes, and the transition to a distributed infrastructure is done in days, not months. It's all part of the approachsingle full-service agency: infrastructure, development, SEO and advertising share the same responsibility.
Want to understand if your current hosting is meeting your business needs? We will conduct an infrastructure audit (speed, stability, Core Web Vitals, TCO) and show whether you need to switch to VPS - within 5 working days.
Frequently asked questions
FAQ - what customers most often ask
What if I have a very small website - a landing page without traffic?
For a landing page without advertising and without SEO tasks, shared or even free hosting is an absolutely normal choice. If a site is launched âfor showâ or as a test landing site without an advertising budget, there is no point in paying for a separate VPS. But as soon as the advertising budget goes to this landing page - even 500-1000 BYN/month - shared becomes unprofitable: the savings on hosting do not compensate for the losses on expensive clicks with a poor Landing Page Experience. This is where the line draws: if you have an advertising budget, you need a VPS.
What to choose if the budget is really very limited?
Modern cloud VPS is already more affordable than you might think. For Hetzner Cloud, DigitalOcean, Vultr, BunnyCDN, and a number of Belarusian and Russian providers, basic configurations start from 4â8 USD/month, which is approximately 15â25 BYN. This is less than the self-respecting âadvanced shared tariffâ, and is absolutely affordable even with the most modest marketing budget. If itâs really difficult, start with a basic VPS (15â25 BYN/month), and not with a shared one: itâs not more expensive, but itâs an order of magnitude better quality.
How do you know when it's time to increase VPS resources?
Pay attention to three metrics. The first is the average CPU load: if it is consistently above 60â70% under normal load, or sharp spikes up to 100% during advertising peaks, itâs time for more cores. The second is RAM usage: if it is constantly above 80% or swap is triggered frequently, there is not enough memory. Third, TTFB and LCP in real Chrome User Experience (CrUX) or Search Console data: if they are creeping up, this is an early signal of resource shortage. On modern cloud VPS, vertical scaling (add CPU/RAM) is done in minutes - this is one of the main advantages over shared.
VPS vs cloud solutions (AWS, GCP, Azure) - what to choose?
For an average commercial site (up to 100 thousand sessions/month), the classic cloud VPS from Hetzner / DigitalOcean / Vultr is almost always more profitable than hyperscalers. AWS/GCP/Azure is not only servers, it is a huge ecosystem of managed services, and for simple tasks their price tag is usually 2-3 times higher for comparable resources. Choosing AWS/GCP/Azure is justified when you need specific managed services (large databases, data analysis, machine learning), complex multi-regional architecture, or integration with corporate systems. For a typical commercial site this is unnecessary.
Managed VPS - does it make sense?
Managed VPS is a VPS for which the provider takes over administration: updates, basic security, backups, monitoring, and sometimes stack configuration. For a business that does not have its own DevOps/administrator and does not have an infrastructure contractor, this is a reasonable choice: an additional payment of 20â40 BYN/month to the VPS price removes all technical concerns. The downside is less flexibility: you can rarely deliver something non-standard. The golden mean is to order a VPS from a good provider, and entrust the administration to a specialized infrastructure contractor (this could be the agency itself that makes the site).
Do I need a CDN if I have a good VPS?
Yes, it is needed - and it is almost always free or almost free. Cloudflare's free plan covers 80% of the average website's needs: DDoS protection, global static cache, SSL, acceleration on all continents. Even with an ideal VPS in Minsk or Warsaw, a CDN will speed up the site for users from Poland, Russia, the EU and the USA - simply due to the nearest points of presence. The VPS + Cloudflare combination is the standard for infrastructure quality in 2026, and it is free or almost free.[5]
Iâve been on shared for several years now - when and how to move?
The answer is simple: once a site receives an advertising budget or active SEO goals, the move will pay for itself very quickly. The move itself requires 1â3 business days of specialist work: setting up a VPS, deploying a stack, migrating files and databases, setting up DNS, reconfiguring external services (mail, analytics, CDN). The move is done in test mode on staging, DNS switching is done in a working window with minimal downtime. If the URL structure also changes, this is a separate project, see the material aboutsafe SEO migration.
Conclusion
The right hosting choice depends less on abstract preferences and more on how much instability the business can afford.
Once traffic, integrations, and revenue depend on the site, infrastructure should be chosen like operating capacity, not like a commodity checkbox.
It's worth checking yourself with a simple test: ask if your current hosting can handle tripling your traffic next week - without TTFB drawdown, without problems with Core Web Vitals, without calls to support. If the answer is âyes, it can handle itâ, it means you are already on a VPS or managed cloud. If the answer is âI donât know, probably not,â this is the best time to upgrade to a VPS before your traffic triples unexpectedly. In 2026, it's not difficult, it's not expensive, and it's the cheapest lost claims insurance you can buy for a commercial site.
Sources
- web.dev - Core Web Vitalsâ official Google documentation on a set of metrics (LCP, INP, CLS) used in ranking through Page Experience. TTFB and server stability are direct components of LCP, which directly links hosting type to search rankings.
- web.dev - Time to First Byte (TTFB)â Google's official guide to TTFB: how it is measured, what affects it, why its quality is directly determined by hosting capabilities. The central metric when comparing shared and VPS.
- Google Search Central - Page Experience- Google Search documentation on Page Experience as a ranking signal: links Core Web Vitals, mobile friendliness, HTTPS and other site quality attributes directly to ranking positions.
- Google Ads Help - Landing Page Experienceâ an official description of how Google Ads takes into account landing page quality (speed, stability, relevance) in the Landing Page Experience, one of the key factors of Quality Score and cost per click.
- Cloudflare - What Is Caching?â Cloudflare's introductory guide to caching: why different cache levels are needed (browser, CDN, origin), how they interact, and why a fast origin server (i.e., a good VPS) is critical even with a CDN.
- HTTP Archive Web Almanac 2024 - Performance- annual report on the evolution of web performance based on data from millions of sites: statistics on TTFB, LCP, INP and CLS for different types of hosting, industries and site categories - an industry benchmark when assessing your infrastructure.
