Service Level Agreement

Introduction

We don’t think you should have to consider the procedure for what goes wrong when you choose a hosting provider — our reputation ought to precede us. However, over the years we have had some demand to create a firmer Service Level Agreement, and these terms represent a clarification of our existing practice — the way we put things right if they go wrong.

Definitions

  1. “Bytemark”, “we”, “Us”, “our” means Bytemark Limited, Novus Business Park, Opus Avenue, York YO26 6BL, trading as “Bytemark Hosting”.
  2. “Service” means any hosting or connectivity service offered by Bytemark, including but not limited to Virtual Machines, BigV/Cloud Servers, Dedicated Hosts or ADSL connectivity.
  3. “Customer”, “you”, “your” means you, the person or persons in whose name(s) a service is maintained by Bytemark.
  4. “Technical contact” means a person or persons nominated by the Customer with whom Bytemark can discuss all matters of a technical nature relating to the Service
  5. “Network” means the combined functionality of all routing, switching, cabling equipment and interconnect agreements under Bytemark’s ownership or control
  6. “Bootable” in the context of a hosting service means that the host in question has the ability to start up and run its normal functions — hosting web sites, receiving email or whatever other function the service has been set up to provide for the customer.
  7. “Working hours” means 9.00am to 5.00pm, Monday to Friday, excluding English bank holidays (see the BERR web site for the full list)

Service Level Agreement (standard)

This document describes the standard level of service that all hosting customers can expect from Bytemark. It attempts to quantify the levels of service that all hosting customers can expect, and the remedies we offer if we fail to provide service at those levels.

There are absolutely no circumstances in which Bytemark will:

  1. Guarantee a remedy for a single failure to meet this service-level agreement that exceed either a refund of 1/12th of a fee paid for annual hosting for an affected service, or a full monthly fee if the service is paid for monthly. This is referred to a as the “maximum monthly refund”
  2. Accept responsibility for consequential damages as a failure to meet the terms of this Service Level Agreement. It is the customer’s responsibility to ensure that the consequences of any such failure are foreseen, and a suitable backup plan in place where there may be financial consequences of a failure. Bytemark will always advise the customer on contingencies to the best of our abilities.

Unless stated in a supplementary service level agreement, Bytemark offers no guarantees on the correct functioning of software installed on any hosting product, and it is entirely the customer’s responsibility to maintain such software. This includes all standard security updates.

Bytemark temporarily suspends all guaranteed service levels for customers who have not settled invoices within the agreed terms.

Core network

You can expect our core network to be available 100% of the time — that is to say that our infrastructure will provide two-way traffic from any other properly-routed internet IPv4 address, to the IPv4 addresses allocated to your hosting product(s). ADSL tails are not considered part of our core network, and are subject to a separate part of this SLA (given below).

We acknowledge that anything less than 100% is a lapse in the service level that you expect. On request we will refund 10% of the maximum monthly refund for every full 30-minute period that the network connectivity falls below this figure, accumulated during any consecutive 30-day period.

We exclude from this guarantee lapses in service where Bytemark cannot communicate with IPv4 addresses from other organisations that are not properly routed at the time of the lapse; the following causes are some examples of where some global IPv4 addresses will end up as “improperly-routed” and excluded from our guarantee:

  1. another organisation misconfiguring its peering arrangements with Bytemark, causing its IPv4 addresses to become unroutable as a result
  2. failure or major changes in routing by a global “tier 1” transit provider, even where Bytemark are not direct customers of said provider
  3. misconfiguration of a hosting product by the customer (as detailed below)

No guaranteed service level is offered for connectivity to IPv6 addresses, as we do not feel that the global IPv6 network is reliable enough to offer such guarantees.

Hosting products (general)

All our guarantees on hosting are conditional on the customer’s hosting product being properly configured and maintained. Specifically this means that:

  1. Bytemark must have a Technical Contact available, either the customer or another party nominated by the customer to make minor configuration changes at our request which may be necessary to facilitate smooth running of our network
  2. your hosting product should not be “thrashing” (i.e. the system constantly using swap during normal operation)
  3. your hosting product should be completely under your control, and not have been compromised by a third party
  4. your hosting product should not be blocked from responding to ICMP pings over its network interface

If any problems arise with your hosting as a result of your hosting product not being maintained correctly, Bytemark cannot guarantee any particular availability or performance.

We may also need to perform maintenance on your hosting which we consider essential to the ongoing reliability of our network or hosting platforms, which we will attempt to warn you about at least 7 days in advance. Such maintenance is excluded from any guarantees of uptime where the customer has been warned 7 days in advance.

Hosting products (Cloud Servers)

These conditions are specific to hosting on our Cloud Servers platform.

You can expect your Cloud Servers to be available for 100% of the time — that is to say that once you start a virtual machine running, it won’t stop running due to hardware faults.

If there is a failure of a Cloud Server, providing the customer’s machine was “properly configured” at the time of a failure, Bytemark will refund 10% of hosting fees for every full hour that it is down or unreachable. We define “down or unreachable” as either the system not running at all, or running slowly enough that a static web fetch of 5KB or less, or a one-line SSH banner return takes more than 20 seconds to return.

You can also expect that the Cloud Servers API will be available 100% of the time, that is to say you should be able to issue requests to create, delete or alter the specification of virtual machines at any time.

If there is a failure of the Cloud Servers API, we will refund 10% of your Cloud Server fees for every full hour that the service remains unavailable, or returning server errors.

Hosting products (virtual machine)

These conditions are specific to hosting on our Virtual Machine platform.

You can expect our virtual machine platform to be available for 100% of the time — that is to say that once running, it won’t stop running due to hardware faults.

However we may occasionally need to perform maintenance that will involve stopping or moving customers data, which can incur 30-90 minutes of down time for each virtual machine. Bytemark will endeavour to warn customers of such outages up to 3 days in advance, but the customer must acknowledge that this is not always possible in the event of sudden hardware problems.

If there is a failure of the service, providing the customer’s VM was “properly configured” at the time of a failure, Bytemark will refund 10% of hosting fees for every full hour that it is down or unreachable. We define “down or unreachable” as either the system not running at all, or running slowly enough that a static web fetch of 5KB or less, or a one-line SSH banner return takes more than 20 seconds to return.

Hosting products (dedicated server)

These conditions apply to customers who rent a dedicated server from Bytemark.

You can expect your dedicated host to be available for 100% of the time unless it has suffered a hardware failure. If your host suffers a hardware failure which brings down your hosting completely, you can expect Bytemark to repair or replace the faulty part within 4 working hours of your reporting it to us, or 24 hours at the outset if the failure occurs outside of 9am-5pm Monday to Friday.

If the hardware failure is not remedied within the stated times, we will refund 10% of your maximum monthly refund for every hour or part-hour that the host is not in a bootable state, up to a maximum of 100%.

Bytemark endeavour to spot hardware failures through various external checks on your server, but we rely on you to inform us if you encounter problems, and will discuss any potential problems you might spot through our normal support channels. We do not consider the 4 hour (or 24 hour) replacement period to have started until the fault has been correctly identified by Bytemark, or reported by the customer and subsequently acknowledge by Bytemark.

Service Level Agreement (Plesk)

This supplementary agreement applies to customers who pay a premium for the Plesk control panel on their hosting product, for whom Bytemark will offer some limited assistance in installing and maintaining it.

Bytemark expects that customers with Plesk will only control their hosting product through the Plesk interface, and not use the supplied root login details to make any significant alterations to the software setup. Any modifications made to a system running Plesk are at a customer’s own risk and may impact or complete remove Bytemark’s ability to provide software support.

Within these exceptions, Bytemark guarantee to do the following on the customer’s behalf:

  • maintain the correct licensing status such that the server continues to function for all future versions of Plesk;
  • advise on, and assist the customer in performing major operations such as upgrades and migrations
  • up to once every three months, perform some minor maintenance on the customer’s behalf which are not covered by the Plesk interface such as:
    • editing php.ini
    • clearing a mail queue
    • fixing DNS inconsistencies
    • applying system-level security updates
    • fix any minor setup issues caused by the Plesk software itself

Customers using Plesk remain completely responsible for the software that they install, and must take steps to ensure the security of their server when installing any new application on their server.

Bytemark will assist with any configuration or setup problems with Plesk, and may occasionally choose to ask the vendor to help with a particular problem where we suspect a fault with the Plesk software itself. We rely on the vendor to solve such problems for us, and where this proves necessary the customer may need to wait 3-4 working days to see a resolution to a particular problem.

Bytemark Cloud used to be called "BigV"—nothing has changed except the name! We’re hiring! Please visit careers.bytemark.co.uk to find out more.