This post is designed to deliver helpful information related to optimising your infrastructure. More specifically, this content will help network and server managers optimise IT infrastructure for network printing.
Today, print management is undergoing a revolution. Historically focal areas have been centred on reducing the “cost per page” metric. Examples include applying authentication at printers/MFPs, reducing print volumes and utilising cost effective printers, and eliminating excessive high cost devices. This approach only addresses part of the issue.
If you expand the scope of the problem, you will find that IT organisations continue to wrestle with increasing costs associated with the explosive growth of data, including supporting ever growing print streams and document rendering activities. There are opportunities to optimise computing, storage, and network assets while ensuring users printing needs are met.
This series will explore the most common way network printing is deployed in large organisations. In addition, we will take a look at alternative solutions that will lower TCO and optimise CPU and network resources, while maintaining user performance and security requirements. A strong partnership between the printer and server management teams can yield significant benefits for users, as well as position IT as a value- add partner to business units.
How Moore’s Law affects network printing
The growth of data is outpacing the decline in processor, networking, and storage costs historically brought about by innovation.
Businesses are struggling to scale up their networks and data centres to keep pace with the explosive growth of data. Until now advances in network, processor and storage technologies have enabled businesses to absorb increases in data volumes with relatively inexpensive infrastructure upgrades. However, the rate of technological advances in data processing, storage and compression is slowing while the growth in data continues to accelerate.
Moore’s Law, the well-known adage in the semi-conductor industry where, in effect, the number of transistors that can be placed on an integrated circuit doubles approximately every two years increasing computational power or performance exponentially without diminishing returns, has been documented as slowing down.
Yet silicon is approaching its limits for transistor density. Soon CPU manufacturers will need to deliver compute power using next-generation technology, such as quantum or molecular computing, which will inevitably make computing, storage, and networking technology relatively more expensive for a given amount of processing power.
As a result, the growth of data is outpacing the decline in processor, networking, and storage costs historically brought about by innovation. Therefore, businesses can no longer count on an ever-cheaper supply of network bandwidth and server CPU cycles to handle an ever-expanding volume of data.
Now businesses must look for more economical solutions that decrease the demands placed on networks and data centre servers. Network printing is one area that places very high demands on networks and servers. New approaches to organising network printing components have emerged that eliminate the demand for server CPU cycles, that eliminate print traffic on WANs and that cut LAN print stream bandwidth load in half.
The most common network printing architecture
It also worth taking a closer look at the most common network printing architectures, what roles the print servers play, and explain how print streams affect WAN and LAN network traffic and print performance.
Traditional network printing over a LAN or WAN requires print servers. In a large office environment with thousands of users, there are often dozens of print servers deployed within the corporate LAN infrastructure. When a branch office or retail store is considered, a Wide Area Network (WAN) connects the main office to the branch offices. Within the branch office, LANs support the local network traffic.
Since network printing streams can take up large amounts of bandwidth, print servers are often placed within the branch office to optimise WAN bandwidth, which is usually considerably lower speed and more expensive than LAN bandwidth. Occasionally, print servers are located in the corporate IT centre to send print jobs to remote printers across the WAN. While this type of deployment provides central management, performance issues such as latency can make this an unacceptable option to users.
The role of the print server
The primary role of the print server is to be a traffic cop. It manages where the print jobs go, as well as queueing to ensure print jobs are prioritised appropriately. Print jobs from desktop clients are sent to an assigned print server, along with the proper “directions” for the specific print job and target printer. Print drivers are often maintained on the print server, and then downloaded by users on individual desktops.
If a print server is deployed within the LAN to support remote printers, it must send print streams across the WAN to the remote printer in the branch office. Depending on user requirements, this may not be a cost effective use of WAN bandwidth, causing many IT organisations deploy print servers locally within the branch office where it utilises the LAN to distribute print streams.
When you look how documents are printed and routed to the network printer, the desktop application initiates a printing request and sends the request and print stream to the print server, which in turn puts it in a queue and assigns it to the proper printers queue. The print server uses the “directions” dictated by the desktop print driver for the target printer. Print data streams can vary in size from very small (small page counts with minimal formatting) to very large (hundreds of pages with embedded images). Deploying print servers to serve users’ needs should be carefully architected based on printing requirements and planned bandwidth requirements.
WAN bandwidth is much smaller and more expensive, which makes it a poor choice for sending large scale print streams from a print server housed in a data centre to a large scale printer at a remote branch office. In this case, it is better to deploy the print server locally in the branch and service the users across the LAN, which optimises less expensive ample bandwidth. This way the users get the performance they need using cost-effective network bandwidth. What happens when you have lots of branch offices with small numbers of desktop PCs, such as bank branches, or retail stores?
Total cost of ownership
IT departments are always trying to lower the total cost of ownership (TCO) of their hardware and software assets. TCO includes not only the acquisition price, but also the maintenance of both the hardware and software that resides on it. In addition, per unit support costs (i.e., help desk) can also be applied to desktop PCs and servers, as well as each unit share of power, cooling, and heating costs. According to Gartner, the average annual TCO for a server is estimated to be £4,300.
When you take into account the average annual TCO costs of servers, this presents an interesting dilemma to IT managers. Is it cost efficient to deploy network printing using the traditional print server architecture in these types of environments?
In future articles, we will explore an emerging alternative to the traditional print server architecture that optimises less expensive computing resources as well as network bandwidth.