Information Technology Reference
In-Depth Information
resources (compute resources, storage, etc) and network performance (latency,
throughput)” (Padgett et al. 2005), the typical minimum content of the SLA should
be the following:
1
. Availability
: this clause indicates the percentage of time, usually on a monthly
basis, in which the Grid/Cloud service supplied by the provider will be available.
With this regard, it is very important to point out that Grid computing is expected
to increase the quality of the services delivered and therefore the customer has
many good reasons to require availability very close to 100 % (the same applies
to Cloud computing). In our example, it is realistic to imagine that
SaaSforyou
chooses
SuperICTResources
as technology provider because the latter is able to
offer an extremely high availability. In the case of a SLA specifically negotiated
by the parties, the customer may be in the position to bargain and obtain a favour-
able and realistic level of availability. While in the case of a standard SLA drafted
unilaterally by the provider the client can only accept or refuse the offer, i.e. he
can enter or not enter into the agreement. The business practice shows that big
international Grid and Cloud computing providers normally guarantee an avail-
ability ranging from 99,9 % to 100 %, and this demonstrates clearly that Grid and
Cloud computing has a notable impact on the QoS to which the provider commits
himself.
2
. Performance
: the objective of this provision is to assure the achievement of com-
monly accepted computing, storage, and network element performance capabili-
ties according to the class of hardware and bandwidths installed. Legally speak-
ing, the content of this clause will depend on the infrastructure adopted by the
provider and therefore the margin for negotiations, especially if the customer is
an SME or a private user (like, we assume,
SaaSforyou
), is usually quite limited.
3.
Downtime and service suspension
: this clause should not find room in an SLA
(or other contract) in a Grid/Cloud environment, and in general in dispersed com-
pute resources scenarios, provided that failures at the level of a single server or
cluster (i.e. Grid/Cloud component) should be compensated by the other ones.
Nevertheless, agreements unilaterally drafted by big international providers of-
ten state that access to and use of the service, or part of the service, may be
suspended for the duration of (i) any unscheduled downtime or unavailability of
a portion or all of the service and of (ii) scheduled downtime for maintenance or
modifications to the service.
4.
Security
: this part of the SLA is of fundamental importance as it will commit the
provider to a certain level of security in order to protect the information and data
supplied by the customer and to prevent harmful components from being deliv-
ered to the customers' computers. The client should therefore pay great attention
to this clause and, if the SLA is negotiated by the parties, should require that
the security standards are set in the contract, so that the provider will be bound
to respect them (see
infra
for further details). The business practice shows that
the SLAs unilaterally drafted by the big Grid/Cloud players tend not to mention
security requirements, so that the customer basically has to trust the supplier.
Provided that trust, as pointed out above, is not a legal category, it is highly ad-
