|
A White Paper answering key questions affecting performance on a Remote Server.
What are the factors that affect speed of system for an ERP user and how ?
In an ERP, there are following components that process and distribute data.
- Database server – it is the computer that stores the central shared data
- Database management system – it is the software that manages database, or acts as storekeeper for the central data.
- Business application – it is the software or set of programs that handle the business logic or ‘Application’.
- Application server – in certain cases, there can be a separate dedicated computer that ‘Runs’ the application or the ‘Business Logic’.
- Client machines or nodes – it is the computer on which end user works. There are typically multiple end users, and accordingly, there are multiple client machines.
- Network components – they are hardware, software and communication services that connect the servers and nodes.
- Network bandwidth – it is the average amount of data that can be transferred between node and server per second.
Accordingly, factors affecting speed or performance of the system for ERP users are:
- Configuration of database server, application server, and nodes – a processor of faster speed and higher RAM and possibly facility of multiple processors improve performance of a computer, and it applies to servers and nodes.
- Capability and tuning of the database management system – while modern database management systems like Oracle and MS-SQL are technically sound, they need a careful tuning of various parameters that is dependent upon statistical analysis of database. This tuning highly affects systems performance, ant it is wise to ensure optimal tuning at least once in six months for best performance. Creation of appropriate indices (index objects that help faster search of records) is also important.
- A business application tested for performance –complexities of business logic that are coded in ERP affect system performance. A bad logic, sometimes involving redundant checks and unnecessary validations mean extra processing and slow performance. A clean and optimized code that is tested for performance and not only for functionality gives superior performance.
- Network components – network components should be secure and complete to ensure perfect connectivity. A loss in physical connection or broken links in communication mean user(s) getting disconnected from the network and from database.
- Network bandwidth – higher the bandwidth, better is the speed. So also, performance can bedegraded if there is a high volume of data exchange between nodes and server. This can happen because of a large number of users working simultaneously exchanging lot of volumes of data with server.
- Transaction volumes and data processing volumes – larger the number of transactions and database updates executed per hour, slower is the performance for individual end-user.
Go to Top
Why systems performance is slow over leased line or over broadband as against that over a local area network (LAN) ?
System performance is slow when a node accesses data from server over a leased line or broadband (as against over a local area network) simply because a leased line or broadband gives a bandwidth of typically lesser than 1 MBPS and LAN with CAT 6 cable gives a bandwidth of 1000 MBPS (million bytes per second). Thus a LAN is thousand times faster than an average leased line connection.
Go to Top
What discipline should one follow while working over a remotely located server ?
Understanding that one is working over lower bandwidths while one works on a remote server, one should observe the discipline as below:
Avoid executing ‘Queries’ that fetch a large number of records from server. A ‘Blank’ query matching all records would fetch all logical records from database server to node. This causes heavy traffic over the communication line, and that drops down performance of all the users. Instead, specify ‘as close as possible’ selection criteria so that only essential records are transmitted by server to node.
Avoid opening multiple sessions with the server or opening of multiple application windows. This again causes unnecessary overheads on the system.
Run only few and needed applications on the node. Close applications that are not needed while working on ERP. Exit from ERP or any other applications when node is not to be in use for some time. Continuously active applications like messaging, anti-virus software, virtual networks and so on too affect system speed.
Slot the timings. Run intensive applications that take up time only at ‘Low load’ part of working hours and with that too with a warning to other users. This is because processes like ‘Stock valuation’ or ‘Material Requirement Planning' (MRP) consume a lot of server and network resources. They cause slower running of all other processes, and at best, should be executed at ‘Off hours’ or ‘Low load hours’ only. It applies to ‘Very large reports’ that run into a number of pages.
Save intermittently when you are creating a transaction with many details. While making a purchase order containing 250 items, it is extremely frustrating if connectivity or communication with server is lost after entry of 240 items. A better approach would be ‘Saving’ and ‘Editing’ the same transaction after each 20 entries. With this habit, in case of trouble, there will be loss of at worst 20 entries and not of 240 entries.
|
