2.2 …TO EFFECTIVELY MANAGE THE ENTERPRISE GROWTH
Effectively managing a growing enterprise requires forethought and planning the design of network and systems management systems. Improper planning of required resource requirements leads to an inability to manage networks and systems as quickly as management growth. How large the network is or how many nodes there are to "manage" is only part of the equation. "What" and "how much" is to be managed within the network and systems must also be entered into the equation. And "who" will implement and maintain it must be calculated as well. This means planning for resources that will enable the management system to go beyond managing the up or down status of individual systems as a whole and moving to managing individual processes and services. It means planning for system availability. These are all part of the requirements gathering process. Each must be an integral part of the deployment plan.2.2.1 Lack of Hardware and Software Resources
There must be a plan for growth, high availability, speed, fault tolerance, and system upgrades. What is managed by the network and systems management system today will definitely not be what is managed next week, next month, or one or two years from now. More users, networks, nodes, applications, and services will be managed than originally planned. Not planning for growth can leave operators with sluggish consoles, cause Network Management System (NMS) polling to slow, or leave little storage for data collection. Overall, it reduces the ability of the system to immediately increase management upon demand.When other organizations within the company learn that there is an organization effectively and proactively managing the network, they may to want to mimic it or piggyback onto that current system to save money. They might ask you to manage their systems. They will want to create a Service Level Agreement (SLA) with you and determine the resiliency of your system.What are your planned outages? What is the MTBF on your systems? What is the support contract on your systems24x7? Are there fail-over systems in place? Are there enough resources to manage their systems as well? These factors must be an integral part of the plan.Most organizations don't plan for future hardware and software upgrades and how the upgrades will affect customers. Purchasing systems that are easily upgradeable by processor or backplane swap will prevent rebuilding everything from scratch and restoring data. Upgrading hardware and software will be considered an outage; possibly several hours of outage. The ability to minimize an outage of the NMS should be in the overall design of the NMS system. Fail-over to another system in order to maintain the customer's SLA can necessitate the purchase of additional hardware and software in order to provide maximum uptime to the customers.If possible, budget for development or "crash and burn" suites in order to test everything before placing it into production. Use the suites to test software upgrades, to fully test any procedure, and to avoid longer-than-planned outages, and even avoid having to back out of an upgrade. Implementing anything on production systems is precarious. One incorrect command can bring the testing of the recovery procedure to fruition.2.2.2 Lack of Human Resources
Management products such as the OpenView suite of products don't behave as expected out-of- the-box. They are fully customizable. One can't simply install them, discover the network, deploy a few agents, magically generate a help desk ticket, page personnel fix the problem within the appropriate SLA, and automatically close the trouble ticket automatically when an alarm is cleared. It takes trained, knowledgeable personnel, and headcount to use and integrate these products.Companies may have, at most, one to three people managing, developing, configuring, and deploying suites of OpenView products at multiple sites across the country. They perform all the design and implementation of each product. They are also responsible for being the system administrator (NT and UNIX), the database administrator (of many database flavors), troubleshooting, and backing up each system. Yet often, a year later, management asks why the entire enterprise is not being managed with all the "bells and whistles" they promised their customers. It's not the employees' fault; it's a lack of management expertise in the requirements to deploy, to deliver what was promised, and to know and hire the proper amount of people with the appropriate talents. It generally leads to a contract being made to a company that is experienced in OpenView product delivery.The company forms a team with the appropriate skill sets to get the project on track and completed. It gathers the functional requirements and compiles a project plan and presents it to the customer. The project plan shows the steps and personnel required to complete the plan on time. The personnel most likely include a project manager, solutions architect, and a number of consultants knowledgeable in each product necessary to meet the functional requirements and for knowledge transfer to the customer to complete the project. A deployment team is also created to deploy the systems to multiple sites. Eventually, the project is completed and it is time for the knowledge transfer to those who will be managing the network and systems management system. How many people are provided who are properly trained in the products for which they will be administering? The answer to that question is usually "zero." The person or persons supplied may have had a class on one product, but have not been to a class for all the products. At this time, there isn't the breadth of knowledge about how each product works and integrates with another. It is nearly impossible communicate all the knowledge provided by skilled OpenView consultants and to a couple people before the project completion who can then train others on how the products and systems are configured and how they are integrated.How many people should there be? That will depend on the number of different products, operating systems, databases, services, etc., at each site and on the terms of the SLAs. The term "products" should include the operating system, backup products, and so on. There is no magical number, formula, or algorithm for an exact number of people required. More often than not, there are too few people on the team with the appropriate skill levels to manage and deploy systems across large environments. Administrators of networks and systems that are going to be managed need to be involved in the various aspects of network and systems management. These aspects include what services need to be monitored, how they are monitored, when to alarm, when and who to page, and so forth. Get them involved.2.2.3 Inadequate Training
Far too often, those who configure the products used within a particular network and systems management system are inadequately trained. They have some training on some products, but not enough to adequately deploy and maintain large environments. The more training that can be given to the personnel, the better it will be for them, the company, and the health and welfare of the NMS. Personnel who are given training classes will have a better understanding of each products capabilities and integration points. They will be able to architect and design a solution faster than they will from months of on-the-job-training. It boosts their morale and may make them less fearful to take on more responsibility as it pertains to the NMS systems and the systems they are managing.While in class, they can network and discuss issues or possible solutions with the instructor and classmates who may have already resolved a similar problem. Many of these students have years more experience in network and systems management. Classroom training can prevent one from writing scripts for which commands already exist to provide the required functionality.A common fear from management is that one will take the training and leave to find a better position for more money or responsibility at another company; thus, the lack of money most companies are willing to spend on training. It is generally junior personnel, who lack experience, who tend to jump ship after training and some months of experience. The technical interviewer at the company should detect a lack of experience, but the interviewer is not always technically sound. To avoid having employees leave the company after paying for training or certification, many companies have instituted policies that require a specific length of stay.It is highly recommended that before attending an OVO UNIX training class that you first attend NNM training. OVO UNIX does not work without NNM! If any SNMP events (traps, alarms), especially Node Up, Node Down, Interface Up, Interface Down, etc., are to be seen in the OVO Message Browser, they will come through the NNM Event system. The NNM training class covers configuration of these events. Any installation of OVO UNIX requires a complete installation of NNM. NNM's processes run, discovery runs, traps are received, and some data collection occurs. Any solid OVO UNIX implementation requires a solid NNM implementation. In order to have a solid network and systems management system that encompasses both NNM and OVO UNIX, both classes should be taken.If a company does not provide adequate training, find some practice time. Find a system, or systems, on which evaluation software can be installed and find out how it works there before implementing on production systems. The vast majority of all OpenView products come with an "instant on" license that is good for 60 days. Test all the theories and ideas before putting them into production. When not busy…practice! Learn all that can be learned about the products and how they function without worrying about breaking anything on a production system. |
لطفا منتظر باشید ...
