Shared, VPS or Managed Hosting: What You Need

Hosting is the one line in a website budget that almost nobody questions. Someone says ₹3,000 a year, you nod, and it renews quietly until the day the site goes down during a campaign. Then you find out what you bought.

The three options you will be offered are shared hosting, a VPS, and managed hosting. They are not three quality tiers of the same thing. They are three different divisions of labour, and the right one depends on two questions you can answer yourself: how much traffic do you actually get, and what does an hour of downtime cost you?

What you are actually renting

Every website lives on a computer in a data centre. That computer has a processor, memory (RAM), disk space, and a network connection. Hosting is renting a slice of it, plus somebody to keep the building powered and connected.

Shared hosting means your site sits on one machine with dozens or hundreds of others. You get a control panel, some disk space, and a share of the processor. You do not choose the operating system, and you cannot install much.

A VPS — virtual private server — is one physical machine carved into fixed slices by software. Your slice has guaranteed RAM and processor allocation that your neighbours cannot eat. You get root access, meaning full administrative control. You also get full administrative responsibility: security updates, firewall, backups, the lot, unless you pay extra for someone to do it.

Managed hosting is a service wrapped around either of the above. The hardware matters less than the contract: somebody else handles updates, backups, security patching, caching configuration, and answers the phone when things break. You are buying labour and accountability, not machinery.

That is the honest hierarchy. Shared is cheap because you share a processor. VPS is dearer because you do not. Managed is dearer still because a human is on the hook.

The word "unlimited" is doing a lot of work

Shared hosting is sold on storage and bandwidth — "100 GB space, unlimited bandwidth". Neither is usually what stops your site.

What stops your site is the CPU and memory limit buried in the fair-use policy. A typical shared plan caps how many processor seconds your account may consume, and how many PHP processes can run at once. A WordPress site with a page builder, a slider, five tracking scripts and an uncached homepage can burn through that on a few dozen simultaneous visitors. When you hit the ceiling, the host does not email you. Requests queue, pages crawl, and some visitors get a 503 error — the HTTP status code meaning the server is temporarily unable to handle the request.

So the useful question for a shared plan is not "how much space" but:

• What is the CPU and entry-process limit, and where do I see my usage?
• Is server-level caching included, or am I relying on a plugin?
• Where is the data centre physically?
• What exactly do the backups cover, how far back, and what does a restore cost?

That last one catches people. Plenty of cheap plans take a weekly backup and charge a restore fee, which you discover at the worst possible moment.

Where your server sits changes what visitors feel

If your customers are in India and your server is in Arizona, every request makes a round trip across the planet before a single pixel appears. That shows up as slow time to first byte — the gap between the browser asking and the server starting to answer — and it drags on everything after it.

Google's Core Web Vitals set a Largest Contentful Paint target of 2.5 seconds or better for a good rating. LCP measures when the biggest visible thing on screen finishes loading. If your server is burning several hundred milliseconds on distance alone, before your theme has even run a database query, you have handed away a chunk of that budget for nothing.

Indian data centres are common now — Mumbai especially, plus Chennai, Delhi NCR and Bengaluru. If your buyers are domestic, host domestically. If you sell to the Gulf, the UK or the US as well, that is what a CDN is for: a content delivery network keeps copies of your images, CSS and JavaScript on servers near the visitor, wherever they are. A CDN does not fix a slow origin server, though. It only distributes the static bits. Server-side slowness still has to be fixed at the server.

Matching the plan to the actual risk

Forget aspirational traffic. Open your analytics and look at last month.

Your situation	What fits	What you are paying for
Brochure site, a few hundred visits a month, enquiries by phone and form	Good shared hosting	Somebody keeping the lights on. Nothing more is needed.
Content site or blog, steady few thousand visits, occasional spike from a campaign	Shared with real caching, or entry managed WordPress	Caching and a host who will not throttle you on a good day.
Ecommerce, live payments, carts and logged-in users	Managed, or VPS with a managed layer	Guaranteed resources plus a human on call. Carts cannot be cached.
Custom application, CRM, internal tools, unusual software stack	VPS	Root access and the freedom to install what you need.
Any site where an hour down costs more than a month's hosting	Managed	Response time and accountability, which is the entire product.

Note the ecommerce row. Shopping carts and account pages are personal to each visitor, so page caching cannot serve them from a saved copy. Every one of those requests hits PHP and the database. That is precisely the workload shared hosting's CPU limits are designed to restrain, which is why small shops that outgrow shared hosting usually discover it on their busiest day.

Things in this category I would not buy

Lifetime hosting deals. One payment, hosted forever. The maths does not work for anyone running a data centre. Either the company disappears or the service degrades until you leave. Both outcomes cost you a migration.

An unmanaged VPS for a business with no technical staff. This is the most common expensive mistake. The plan costs a little more than shared hosting and gives you far more power, so it looks like an upgrade. Then nobody applies the security patches, nobody notices the disk filling with logs, and nobody has tested the backup. An unpatched server on the open internet is a liability, not an asset. If you buy a VPS, buy the management alongside it or have someone on retainer for ongoing technical support.

Four-figure monthly "enterprise" hosting for a ten-page brochure site. Some agencies mark up hosting heavily and call it managed. Managed is a real service with real work in it, and it is worth paying for when you need it. Ask what is actually done each month: patching cadence, backup frequency and retention, uptime monitoring, support response window. If those questions get vague answers, you are buying shared hosting with a nicer invoice.

Hosting bundled with SEO or maintenance as one inseparable fee. Bundles make it impossible to see what you are paying for and painful to leave. Keep the hosting line separate so you can price it, judge it, and move it.

Paying for a bigger server to fix a slow site. If your homepage loads eight uncompressed images straight off someone's phone, doubling the RAM changes nothing. Fix the site first. More often than not, the hosting was never the bottleneck — that is usually a build and front-end problem.

The questions that matter more than the tier

Whichever tier you pick, these separate a host you can live with from one you will fight:

1. Who holds the account? It should be in your company's name, with your billing email, and you should have the login. Not the agency's reseller account.
2. How do I get a full copy of everything? Files and database, downloadable by you, today, without asking permission.
3. Is there a staging site? A copy where updates are tested before they touch the live site. This turns "the plugin update broke checkout" from a crisis into a Tuesday.
4. Which PHP version are you running, and how do you handle upgrades? Old PHP versions stop receiving security fixes. A host that leaves you on an unsupported one is not maintaining anything.
5. What is your support channel and typical response time? Ticket-only support with a day's turnaround is fine for a brochure site and unacceptable for a shop.

One more thing worth being honest about: uptime guarantees. A 99.9% SLA sounds absolute, but it permits roughly forty-four minutes of downtime a month, and the compensation is usually a credit against your hosting fee — not against your lost sales. Read it as a statement of intent, not insurance.

What to do this week

Find your hosting renewal email and note three things: the plan name, the renewal price, and whose name the account is in. Then check your analytics for last month's actual sessions and your busiest single day. Most Indian small businesses discover they are on a plan that comfortably fits — and that the real speed problem sits in the site's own code and images, not the server.

If you are running payments, stock or customer logins, the calculation changes, and the cost of an hour offline is the number to work from. If you would like a second opinion on which side of that line you sit, tell us what your site does and roughly how much traffic it gets and we will tell you plainly whether your hosting is the problem. More on builds, speed and systems that run without you on the blog.

Questions we get asked