What Makes for a Good SLA?

Like car insurance or self-assessment, it's easy to see100% SLAs are meaningless and misleading. "In the
the Service Level Agreement (SLA) as a bit of areal world there are outage; no system is perfect.  It
black art; but they're vital in ensuring IT serviceis important for customers to be realistic.  It is then the
providers deliver availability, responsiveness, quality andresponsibility of a good MSP to design a solution that
communication.  Also, with managed services climbingbalances the number of 9s in the availability guarantee
the IT agenda and other disciplines like Software as a(99.9%? 99.99%? 99.999%?) to the number of
Service gaining serious momentum, a watertight SLA£££ the customer is prepared to pay.  An MSP
can make the difference between the success andSLA stating 100% means the customer has no way of
failure of your critical business systems and even theknowing what the ‘real' performance will be."
business itself.5. Defining Downtime
So, what makes for a good SLA?It's vital to have a clear understanding of how the
1. Service Summary or Descriptionservice provider defines ‘downtime'.  Most don't
Should detail the names of the provider and theconsider that upgrades constitute service downtime
customer, along with the obligations the latter has tofor example, so won't compensate you for them.  Pin
fulfil to stay within the bounds of the SLA.  You'lldown details such as how fast the provider will
typically be asked to provide up-to-date contacts,respond to service requests, how long they'll take to
network topologies and escalation paths for example.detect, report and action problems, how long upgrades
Should also list the support level – gold or platinumwill take and so on.
say – you're buying into ie. How fast will the service6. Service Requests
provider respond to service requests?  How manyMost SLAs allow a set number of service and
requests are you permitted during the weekly oremergency requests during each service period and
monthly service period? What's the notificationit's important to understand the difference. Some
process?providers, for example, class any task performed
Most importantly of all, what is your general serviceoutside standard business hours as an emergency,
availability guarantee?which could become an issue if most of your requests
2. Hardwarefall outside the hours of 9 and 5.  Some providers limit
Whether installed onsite or managed remotely, thethe number of your IT personnel permitted to initiate
SLA should state clearly what hardware is to berequests; some define certain tasks as constituting
provided.  Once you're sure of the hardware in use,more than one request; some charge extra for certain
ask more specific questions about spec, performance,service requests.
throughput, size, upgrades, and so on.7. Monitoring & Reporting
3. SoftwareMany providers have substantially improved their
Most providers use products from name-brandprocesses for reporting on metrics such as bandwidth
vendors.  Others use open-source software.  Manyutilisation and uptime in recent years, but capabilities still
use both.  In any event it's important to know whatvary greatly from provider to provider, so ask
software your service is subject to and where,questions.  Is the most up-to-date configuration
particularly if you have special requirements oravailable for review online? How often will you get
stipulations such as bans on unsupported software. reports based on firewall, IDS or VPN logs?  How
Visibility over the software being used will also give anabout ad hoc and custom reporting?
insight into the provider/software vendor relationship. 8. Other Components
For example, if your supplier is provisioning your firewall- Some providers, especially ASPs, source their
using Cisco PIX but has no qualified CCIEs on staff,services from multiple third parties – network
you need to know about it.providers, infrastructure providers, application
4. Service Availabilitymanagement providers – each with their own SLAs
Probably the most important element of thethat may in turn impact yours.  Find out what
agreement, this section should describe precisely whatcomponents of the overall solution are covered under
you're guaranteed under the terms of the SLA,your SLA.
including critical aspects like guaranteed uptime- Ask about guarantees against dropped
percentage.  This is particularly crucial as, while anconnections.  Some SLAs offer money-back for the
uptime figure such as 99.5% looks pretty high at firsttime your connection is down.
glance, it would mean your systems could be down for- Some providers measure network traffic rates
as much as 216 minutes per month without thebased on the packets going in and out but don't allow
provider incurring any penalty.  (You should befor dropped packets.  Push for a guaranteed packet
compensated for any downtime outside this tolerance,loss rate in your SLA.
which is generally a case of the provider not invoicingOverall if the provider won't give key guarantees, look
for the period in question.)for a replacement.  But remember, an SLA has to
But beware the provider offering 100% availability. work for both parties and shouldn't be simply a big
Few do and for good reason; even if it were possiblestick with which to hit the supplier in the event of a
to deliver 100% availability – which is to say theproblem.
least debatable – it would be prohibitively expensiveDefining, negotiating and measuring can be difficult, but
for both provider and customer.  Also, as Martinit's also vital if you want a meaningful SLA.
Saunders, Group Product Manager at Claranet puts it,