Skip to content

Managed server clusters that keep running when a server fails

Load-balanced web servers and replicated databases, built from our dedicated servers in our own data centres and fully managed by our engineers. Add GPU or specialist servers when you need them.

Design it below and send it to us. We quote every cluster for its workload.

Load balancer3 of 3 healthySite up

web-1Healthy

Serving

web-2Healthy

Replaced

web-3Healthy

Serving

db-1Primary

Live copy to db-2

db-2Replica

Up to date
Requests served: 12,560Fully managed by Tremhost
  1. web-2 failed its health check
  2. load balancer stopped sending traffic to web-2
  3. web-1 and web-3 serving every request
  4. web-2 replaced and back in service
Illustration: when a web server fails, the load balancer routes around it and the site stays up while we replace it.

One server is one point of failure

If your application runs on one machine, its downtime is your downtime. A cluster puts two or more servers behind a load balancer, so one can fail without taking you offline.

8.76 hoursa year down, for one server that is up 99.9% of the time
About 32 secondsa year with both down, for two such servers behind a load balancer

Arithmetic, not a guarantee: (1 − 0.999) × 8,760 hours = 8.76 hours. Both servers down at once, if failures are independent: (1 − 0.999)² × 8,760 hours × 3,600 = about 32 seconds. It leaves out the load balancer, the network and failures that hit both servers together, which is why we design and manage those too.

A server fails

One server: The site is down until the server is repaired or rebuilt

A Tremhost managed cluster: The load balancer stops sending traffic to it; the other servers keep serving while we replace it

Traffic grows

One server: One machine's cores and RAM are the ceiling

A Tremhost managed cluster: Add web servers to the cluster as traffic grows

Your data

One server: One copy of the database on one server

A Tremhost managed cluster: A second database server can keep a live copy

Running it

One server: Load balancing, networking, failover, patching and backups to build and run yourself

A Tremhost managed cluster: Fully managed by our engineers, with daily backups kept 7 days

Maintenance

One server: Patching or upgrading the server means taking the application offline

A Tremhost managed cluster: Servers can be patched one at a time while the others keep serving

Design your cluster

Pick the servers from our dedicated line and the data centre. Send the design to an expert and we'll quote it, including how long it will take to build.

Web and application servers
Database servers
GPU servers and location

Your cluster

  • Load balancer

Web and application

  • web-1
  • web-2

Database

  • db-1
  • db-2
  • Servers4 + load balancer
  • CPU72 cores, 144 threads
  • RAM320 GB
  • Web servers2 × Dedicated 3: AMD Ryzen 9 5900X, 32 GB, 2 × 512 GB NVMe, software RAID
  • Database servers2 × Dedicated 4: AMD EPYC 7413, 128 GB, 2 × 960 GB NVMe, software RAID
Send this design to an expert

Fully managed, with daily backups kept 7 days, DDoS mitigation and 2 IPv4 addresses per server. We reply with a quote and a build timeline.

Servers for special jobs

Add any of these to a cluster, or run them on their own. Specs change as hardware does, so an expert sends you the current options.

  • GPU servers

    For AI training and inference, rendering and scientific computing.

  • High-IOPS NVMe servers

    For heavy databases, analytics and video processing that hit storage hard.

  • In-memory database servers

    Large RAM configurations for databases that keep everything in memory.

  • High-density servers

    For big data platforms such as Hadoop, Cassandra and MongoDB.

Build your cluster in any of our five data centres

Put the cluster close to its users. Johannesburg is the closest to Southern Africa.

Tremhost VPS data centres JohannesburgLimburgRoubaixNew YorkMiamiHarare office
Map of Tremhost's data centres: Johannesburg, New York, Miami, Limburg and Roubaix, with Tremhost's office in Harare.
  • JohannesburgSouth Africa

    Closest to Zimbabwe and the rest of Southern Africa.

  • LimburgGermany

    Central Europe.

  • RoubaixFrance

    Western Europe.

  • New YorkUSA

    East Coast of North America.

  • MiamiUSA

    Southern USA, the Caribbean and Latin America.

Who looks after what

You run your application. We run the cluster under it.

We design the load balancing and failover, build and patch the servers, monitor them, back them up daily and replace anything that fails. You keep root access.

  1. Your application and its dataYou
  2. Load balancing, failover and cluster designTremhost
  3. Patching, monitoring and daily backupsTremhost
  4. Dedicated servers, networking and DDoS mitigationTremhost
  5. Data centres, redundant power and networkTremhost

Included with every managed cluster

Your quote covers the servers you choose. Everything below comes with them.

  • Fully managed

    Our engineers build, patch, monitor and fix every server in the cluster.

  • Daily backups, kept 7 days

    Every server is backed up daily.

  • Load balancing and failover

    Traffic is spread across your web servers and routed around a failed one.

  • Designed networking

    We design the private networking between your servers.

  • DDoS mitigation

    Attack traffic is filtered before it reaches your cluster.

  • Redundant power and network

    Each server has redundant power and network connections.

  • NVMe with power-loss protection

    Fast NVMe storage on every server in our dedicated line.

  • Two IPv4 addresses per server

    Included with every dedicated server.

  • Root access

    The servers are yours to use. We manage them; you keep root.

  • GPU and specialist servers

    Add GPU, high-IOPS, in-memory or high-density servers to the cluster.

  • Five data centres

    Johannesburg, New York, Miami, Limburg and Roubaix.

  • Human support

    Real engineers in Harare, reachable on WhatsApp and by ticket.

Do you need a cluster?

Often one server is enough. Start with the first line that fits.

Trusted by businesses since 2016

4.6out of 5, from 30 Google reviews

Source: Tremhost's Google Business Profile. Read all reviews on Google

These guys have the best support and customer service, they are always available for support, I highly recommend.

Ed Chitambo, Google review

Tremhost provides excellent all-around service as well as speedy customer support. I strongly recommend them.

Russel Jhamba, Google review

Great company for sites. I have been working with these guys for years. Keep up

Promie, Google review
  • CILT Zimbabwe

    Professional institute, Harare

    Membership, events and the WiLAT community on one site that Tremhost has run since 2021.

    Read the story
  • Buildware

    Decorative panels and cabinet hardware, Harare

    A client since 2017: a product catalogue with WhatsApp quotation requests and SEO built around trade terms.

    Read the story
  • Specialist Junior School

    Private primary school, Ruwa

    A school website parents use and pupils publish on, with an events calendar for term dates and fixtures.

    Read the story

Pay in US dollars, the way you already pay

EcoCash, InnBucks, ZimSwitch, Omari, Visa, Mastercard, Bitcoin, Tron or Litecoin. If a payment doesn't go through, message us on WhatsApp.

Cluster questions, answered straight

Still unsure? Talk to an expert or open a ticket.

What is a server cluster?

Several servers that work as one system. Web or application servers sit behind a load balancer, which spreads visitors between them and stops sending traffic to any server that fails. Database servers can keep a live copy of your data on a second machine.

Which servers are clusters built from?

Our current dedicated line, from Dedicated 1 (Intel Xeon E3-1230 v6) to Dedicated 6 (AMD EPYC 9755, 128 cores), all with NVMe storage. GPU, high-IOPS, in-memory and high-density servers can be added; specs on request.

What happens when a server fails?

The load balancer notices that the server is not answering and routes traffic to the others, so the application keeps running. Because the cluster is fully managed, our engineers then repair or replace the failed server.

Is it fully managed?

Yes. We design, build, patch, monitor and back up the cluster. Every server is backed up daily and backups are kept for 7 days. You keep root access.

Where can my cluster run?

In any of our five data centres: Johannesburg, New York, Miami, Limburg or Roubaix. For users in Southern Africa, Johannesburg is the closest.

How long does it take to set up?

It depends on the scope: the number and type of servers, the networking and the software we set up. An expert gives you the timeline with your quote.

Why are there no prices?

Every cluster is designed for its workload, so the price depends on the servers, data centre and services. Use the builder above, send us the design and we will quote it.

Do I need a cluster?

Not always. One busy site runs well on a managed cPanel VPS or a single dedicated server. A cluster earns its keep when an outage costs more than a second server, or when one machine is no longer enough.

Tell us what has to stay up. We'll design the cluster

Send us your application, traffic and data. An engineer will design the cluster, choose the servers and data centre with you, and give you a quote and a build timeline.

Talk to an expert