Re: QoS - Steve your right on Que., I knew it would not take long.
I tried ending this debate under amicable terms over on the
Westell thread stating:
" I'm sure you'll agree that we both had valid view points in our discussion/minor debate."
But you would not let go. With you Firing back with SAM Missiles
"Well, even wanting to be generous, this is STILL not the case.
The specific term "Quality of Service" (note caps) and especially the acronym "QoS", as has been amply clarified for you, has NOTHING to do with YOUR use of it to mean simply "good quality service".
Now you come back with your Digs at me again"
Give it up, rev. It was a debate only in YOUR mind. You continue to be lacking in your understanding of the datacom term QoS. (Hint, it is NOT related to YOUR use of the colloquial, NON-technical descriptions "good service", "quality service", "good old-fashioned service" or "service with a smile") Both Geoff Hollingsworth and I explained this to you.
<..QoS (Quality of Service) In Networking it is how reliable is your End to End Solution to the Client/Customer...>
Didn't you read ANYTHING you were given? Spend some time reading about ATM ("asynchronous transfer mode"). And maybe find someone who knows about networking to help you with this (I'm sure, Geoff could help you) and you can get FAR ahead of your faulty attempts at THIS silly and embarrassing "see-I-was-right" game.
Oh I read it all right, and understood most of it prior to getting into this debate, but that does not mean it is correct!
But even if you eventually ARE successful in finding someone who, like yourself, misuses the term, it is STILL a misuse. You might want to re-read (if you ever DID read it) the article you errantly posted in your own defense, which in fact directly ATTACKED sloppy uses, such as yours, of these networking management terms.
I found someone long ago, but your Greater Than THE Attitude is just to much to for anyone to contend with for more than a few posts.
If the attitude were to not be factored in, your posts on many threads here on SI are informative and usually on target.
But for now I'll let you debate a White Paper on QoS for a while.
Note: My View Point is made perfectly Clear.
Once again I'll try to end the debate.
" I'm sure you'll agree that we both had valid view points in our discussion/minor debate."
Am I still wrong?
Thanks Info Vista Corp.
infovistacorp.com
(snip, snip, snip)
Managing Quality of Service A Strategic White Paper
March 1, 1997
by Richard C. Sturm,
Principal
Enterprise Management Professional Services, Inc.
I. Introduction
Over the last 10-20 years, as global competition has increased, managers have increasingly sought ways to improve their organizations' competitiveness. Often, as a new idea appeared, it was adopted rapidly, and some would say unquestioningly. All too often, these new ideas have ended up merely being short-lived fads that have produced few tangible benefits. Many of these ideas have been held up to ridicule in the popular cartoon, "Dilbert."
Managers of Information Technology organizations (IT) have proven to be susceptible to this same phenomenon. The frequent misuse of popular technology has been rampant. Currently there appears to be the beginning of a groundswell of interest among IT managers in quality of service management (or as some refer to it as service level management). The question must be asked. "Is quality of service management another management fad, or is it a legitimate business focus?" Many cynics would be quick to say "yes." However, a review of today's situation quickly refutes such skepticism.
Traditionally, IT managers have measured the effectiveness of their organization by looking at the various hardware and software components for which they are responsible. The units of mea-sure that were applied have been varied. The most common have been such things as availability,utilization, performance, etc. This has given IT a fragmented, misleading perspective of its effec-tiveness.
It is a perspective that is distorted and misleading. It has led IT managers to believe that they have been doing a excellent job at meeting the needs of their clients for IT services.
In recent years, while IT managers have been congratulating each other on the ever improving quality of service that they have been providing, their clients have had a radically different per-spective.
They have been complaining about the inadequacy of the service and the unresponsive-ness of the IT organization to client requirements. In recent years clients of the IT organization have become more and more dependent upon the services provided by IT. The very existence of many businesses depends upon the delivery of those services. In addition, their clients (users) have become more technically sophisticated. This has given them a better understanding of what they can reasonably expect IT to deliver. Together, these factors have led IT's clients to demand a much higher quality of service from IT. In some cases, it has also led the users to stage a coup, toppling the IT organization from its pedestal and replacing it with an out-sourcer's organiza-tion.
The clients of IT hold the corporate power and control the corporate purse. One way or another, they are determined to receive the level of IT services that they require.
Today's interest in managing the IT organization in terms of the quality of service it delivers has come to the forefront due to the convergence of two major factors. Basically it is the convergence of supply and demand. The confidence of IT managers has been shaken by the complaints of dis-gruntled clients and also by continuing reports of outsourcing deals. IT, and their clients, have a need to accurately measure and report on the quality of the service being delivered by IT. This is the demand side of the equation. True end to end service level measurement has only recently become practical. There are now tools emerging to support the measurement of the quality of the service being delivered by IT.
It must be remembered that not only is IT a service provider, it is also a major consumer of ser-vices.
If those providing services to IT do not deliver a consistent level of service, IT will not be able to be consistent in the quality of services that it delivers to its clients. Those providing ser-vices to IT include groups internal to the company (e.g., data base administrators, application developers, etc.) as well as some that are external (e.g., telephone company). However, it must be realized that managing the quality of service delivered is not limited just to IT organizations.
Certainly, for most organizations the IT function is a critical service provider, but there are many others. For example, the telephone company that provides voice and data links to other compa-nies, is a major service provider. It is only reasonable to demand a guarantee of the quality of ser-vice being provided.
Is the current desire to establish effective measurement and reporting of the quality service being delivered a fad? Absolutely not! As will be seen in this document, quality of service management (service level management) is absolutely essential as a survival strategy for the IT organization.
II. Divergent Views
What has gone wrong? Why have the views of IT and their clients about the quality of service being provided by IT become so divergent? The bottom line answer is that the two groups do not have a common understanding of what is an adequate level of service or even about what the key indicators are of the level of service being provided.
Until now, most IT organizations have not been measuring the quality of the service being deliv-ered to their clients. Instead of measuring the quality of the service being delivered, IT has been measuring various attributes of the hardware and software used to deliver the service to its clients. Certainly this data is valuable for evaluating the performance of the IT employees or planning for adequate equipment capacity in the future. Very technical metrics are also essential in assessing the quality of the service being delivered by the external network service providers (i.e., telephone companies). However, this very technical data is not indicative of the level of ser-vice being delivered by IT to its clients.
There have been several flaws in the strategy employed by IT. First, it has not been possible to capture equivalent information for every component. That is, in examining the end-to-end quali-ty of service being delivered, there are numerous components that impact that service. There are various sources of information about those components and about some portion of the service being delivered. These different data sources do not capture the same types of information, nor is their data capture synchronized. The result is that the data available has been fragmented, scat-tered and disjointed.
Second, for some components, only limited data has been available. In some cases, no data rele-vant to quality of service has been available. The third flaw has been the type of information cap-tured. Certainly, CPU uptime, dropped packets, network congestion, etc. will each impact the quality of the service that is delivered. However, these factors can only indirectly suggest some-thing about the quality of service being delivered. Finally, IT has been struggling under the assumption that the whole (of quality of service management) is equal to the sum of all of the parts. In mathematics, this assumption is valid. In the management of IT services, it is totally false. Consider an simple example in which a typical transaction requires that each of the follow-ing pieces of equipment and software listed be available.
Table 1
Average Minutes of
Component Availability Downtime
LAN 99.97% 0.32
Local Server 99.95% 0.54
Building Hub 99.96% 0.43
Intranet router 99.99% 11.88
Remote Host 99.99% 1.08
Order Entry applic. 99.91% 9.72
Customer data base 99.92% 8.64
Inventory data base 99.96% 4.32
Average/Total 99.66% 36.94
Let's assume that the user group performing order entry demands 99.90% availability for the order entry "function." (Note, the term function is being used to indicate the desired activity, such as entering a new order.) Expressed another way, the client expects to be able to have the function available 24 hours per day, 7 days per week, with no more than 10 minutes of down-time in a week. Looking at the table above, most IT organizations will congratulate themselves on having met the client's requirements in all but one of the categories, and even that was very close to the objective. However, this is probably not correct. Unless there was the coincidence of all of the problems occurring at the same time, then the total time that the function was unavail-able to the order entry department was much greater than the acceptable level. If the outages each occurred at a different time, then the total amount of time that the function will unavail-able to the client was almost 37 minutes! Also, the availability can be calculated by multiplying together the availability of each of the components. The result is 99.66% - far less than the tar-get of 99.99%. In truth, there may be some overlap between the various events. This could reduce the total amount of time that the function was not available to the order entry clerks.
However, overlaps could not conceivably reduce the minutes of outage to the target level.
From the perspective of the end user of the IT services, the availability of individual components simply does not reflect the quality of the service being provided. The one merit of this particular metric is that it does reflect a concept that the clients can relate to, even if it is from a different perspective. If data for individual components is all that is available, then what is needed is a more sophisticated correlation and analysis of the data to translate it into meaningful information.
In terms of managing the quality of service to client organizations (that is, end users), there is absolutely no value in reporting such arcane data points as packets dropped, network latency, repair interval for network outages, etc. With a few rare exceptions, the clients simply do not understand their significance. Even if understood, these pieces of data are not measurements of the quality of the service being provided by the IT organization. It is true that such information may be valuable measurements for internal IT assessments. In fact, some groups develop what are termed "Technical Service Level Agreements." These are basically agreements that are used to look at how well the groups within IT are fulfilling their responsibilities.
Very well, technical measurements are valuable for special purpose assessments of quality, but in their raw form do not reflect the quality of service that is being received by end users. However,. if device availability, as well as other detailed statistics, are not valid indicators of the quality of service being delivered to end users, what is? Before this question can be answered it is necessary to deal with the even more fundamental issue of defining what is meant by the term "quality of service."
III.Defining Quality of Service
To begin with, in considering the meaning of the term "quality of service," it is necessary to assume the perspective of the client - the consumer of the service. The client's perspective is the only perspective that matters in this definition. This is true regardless of the industry. It is as valid in the fast food business as it is in the world of IT. The client is the one utilizing the ser-vice.
The client's perception of the quality of the service being delivered will determine whether or not they are satisfied with that service and whether or not they will want to continue to work with that service provider.
There may be a temptation to consider the adoption of a client's perspective to be biased and possibly even one-sided. However, IT managers must remember that they are both providers and consumers of services. They are the client when dealing with service providers such as the local telephone company. They are also consumers of services from their own organization. The IT department uses the services of various groups within IT to provide the total package of services that they offer to a client. For example, a typical IT department will have a group responsible for operating the corporate Intranet. Another group may be charged with the operation of the cen-tral host computers. Yet the IT department, or the CIO, ultimately uses these services together to provide an end-to-end service package for other departments in the business. In this mode, the IT department is a client for IT services.
From the client perspective the quality of IT services can be thought of in fairly simple terms.
The most important point to remember is that the client is concerned with business functionali-ty.
That is, they need to be able to use the IT service to perform a business function. They see the ability to enter customer orders as an IT service, not as a collection of independent services.
They are not concerned with how the service is provided, or what components make the service possible.
The concerns of the client can be classified into three broad categories. Those categories are listed below:
Availability
Performance
Accuracy
Affordable
A. Availability
Availability focuses upon the question whether the services are able to be used by the client when the client wishes to use them. There is not a magic number of days or hours that provides the right answer for this category. Rather, the objective is to define in advance when the services will be required and meet that definition. There is the most obvious meaning for this facet of quality of service - scheduled hours of operation. Even in this sense, it is important to note that it means actually being able to successfully perform the desired function. All of the required components must be available. Having the family car ready for your spouse to use for a trip to the local mall is an example of this. It is highly unlikely that your spouse will consider the car available for the trip when everything is in perfect running order, but one tire is flat.
There are also less obvious aspects of availability. Consider the case in which a company has a system that allows sales representatives to dial in using their laptop computer to check product availability while they are at a customer's office. In such a case, if a modem port is not available when a sales representative attempts to call, then for that sales representative the service is not available.
B. Performance
Words can hold vastly different meanings for different people. This is certainly true of the word performance. For IT personnel performance will suggest such things as packets per second.
However, for the client, the meaning is simpler - do the IT services function at an acceptable speed? The question of speed may be the question of response time in an online transaction sys-tem.
In another case it might be the time that it takes to move a copy of a file from one office to another, or the time required to load an application to a desktop system from a server.
C. Accuracy
From the perspective of the user of a service, there is a third part to the question of quality. That third part is accuracy. This component is concerned with the question of whether or not the ser-vice performs accurately. An example of this is the question of whether or not e-mail is delivered to the correct recipient. Similarly, in the case of applying transactions to a data base, it is essen-tial that the change is applied to the proper version of the data base. If services do not accurately perform their functions, high availability and high performance are worthless.
Having loosely defined quality of service, let us next look at the question of how to approach the management of that service and the level of the service that is provided. We will defer the ques-tion of which specific indicators or measurements are appropriate to use until later in this docu-ment.
First we must look at how they will be used.
D. Affordable
While availability, performance and accuracy, are strictly related to the quality of service, the cost of the service cannot be ignored. There is a saying in application development, "Fast, cheap, good - you can have two out of three." This expression is true when it comes to delivering any service. There is always a tradeoff between cost of the service and the quality of the finished product, as well as the timeliness of the delivery. It may be possible for a service provider to raise the quality of the service that it is providing dramatically. However, the cost of such an improve-ment might be so great, that it, if implemented, that it would cause the bankruptcy of the cor-poration.
Therefore, while cost is not directly part of the quality of the service being delivered, it is a limiting factor, and cannot be ignored.
IX.Conclusion
The service level agreement is a powerful tool for both service providers and their clients.
Certainly it is possible for an SLA to fail to realize its potential. However, if the guidelines of this document are followed closely, the companies can be very successful with their SLAs.
Managers in service provider organizations can find that the SLA, along with the reports that are produced for it, are very useful tools for demonstrating the effectiveness of their organization, defending it from criticism and also in justifying requests for funding.
In order to be successful in developing and managing an SLA, it is essential that before starting into the process of creating one that managers first address the prerequisites. That is, they must know what data is available (that is, know what they can measure) and they must have SLAs with each of their service providers. They must define the agreements in terms of service indica-tors that are: attainable, measurable, meaningful, controllable and mutually acceptable. They must document their performance with accurate, comprehensive reports that go beyond provid-ing just data and instead deliver information that is derived from a variety of data sources.
Finally, information must be presented in a manner that is relevant to the client organizations.
The technology of network and systems management is finally reaching a stage where meaning-ful service level management is finally possible. At the same time, client groups are becoming increasingly vocal in their demands for higher quality service, and a better return on the money they are spending for a service. This is true of all service providers, whether an IT organization, telephone company, or other. The demand for meaningful service level management and report-ing is going to continue to escalate. Service providers will be wise to prepare now to meet those demands. |
