Alex,

as you know, inside the OGF working group OCCI (stands for Open Cloud
Computing Interface) we are trying to define the guidelines for SLA in
clouds. With this aim I tried to modularize the complex issue of SLA
as-a-whole into a few sub-problems. Per se the SLA topic extends from
the definition of legal details to performance measurement: definitely
too much for a single document that wants to give precise guidelines. So
I started by extracting the resource monitoring sub-problem.

Such functionality is present in many aspects of distributed
computing, and exhibits a sufficiently simple (and sufficiently
understood) interface with other components. In our perspective the
specification of the monitoring infrastructure does not include
neither the indication of relevant metrics, nor the technology used to
retrieve them: we focus on the arrangement of sensors attached to
resources, that may accept as "plugins" specific measurement tools, as
well as measurement aggregators and publishing supports.

In DMTF IS0102 there are sparse references to such a monitoring
infrastructure, and its role is explicitly addressed (for instance in
section 7.8). While reading your document "TR178 Enabling e2e Cloud
SLA Management" the topic is covered in better detail. While defining
the tightly related topic of metrics (in sect.  3.4 of version 0.3),
you notice the importance of defining how metrics are used, and on the
existence of a reliable infrastructure for resource monitoring that
ensures comparable results. In section 4.3 you state that a well
defined service must give the consumer some access to monitoring
facilities, with an interface mediated from a TM Forum document.

The interest for an agreed definition of the monitoring infrastructure
is implicit in a number of future work recommendations in your
document (sect. 6): Recommendation 4 asks the inclusion of the
description of the container of service metrics as a part of the
Service Template, Recommendation 6 indicates the definition of metrics
as a continuous effort, and thus mandates a flexible infrastructure
for their management.

This given, I consider you are probably interested in the working draft
about the resource monitoring infrastructure I have recently proposed
to the OGF-OCCI working group. The purpose of the document, in the
spirit of OCCI, is to define an interface to the resource monitoring
infrastructure. The interface follows a restful approach, and gives
the way to create, configure and discover resource monitoring entities.

I hope that the ideas contained in the document are helpful for your
task, and I will be pleased to receive any comment about the
document (that you find in attachment) from your community.

Faithfully,

Augusto

2012/10/11 Alex Zhdankin <azhdanki@cisco.com>
Submitter's message
Colleagues,

This document was jointly developed and submitted by TM Forum for comments and further input. Please, review this Technical report and provide any comments you may have. The topic of SLA management in Cloud environment is directly related to the work we are doing in several groups, including CMWG, NSMWG and SVPC, so, please take a look and comment.

Thank you and Regards,

Alex.

-- Mr. Alex Zhdankin
Document Name: TR_178_V2012-0921_V0-3.docx
Description
This document adresses the issues of End-to-End SLA management in the Cloud
ecosystem. The goal of this Technical Report is to enable people in
different communities to develop Cloud SLAs and SLA metrics such that
associated systems can be joined to facilitate end-to-end SLA management of
a single cloud or multi-cloud environment. This Technical Report (TR178),
while organized by the TM Forum, takes an outside-in look by reviewing
existing relevant industry work (DMTF, OGF, NIST, ITU-T, ISMA, OASIS and
other), as well as the TM Forum Frameworx, SLA management Handbook and
Cloud SLA Application Notes. It then recommends a set of business
considerations and architecture design principles that are required to
support end-to-end Cloud SLA Management with the aim to facilitate
discussion regarding SLA consistency across Cloud Deployment Models and
Services Models.
Download Latest Revision
Submitter: Mr. Alex Zhdankin
Group: Network Services Management Working Group
Folder: TMF alliance
Date submitted: 2012-10-11 06:10:37



--
Augusto Ciuffoletti
Dipartimento di Informatica
Università di Pisa
56100 - Pisa (Italy)