Networks have become an integral part of most organizations these days. Today more and more employees are accessing the network for Internet, intranet, other network-based applications, or for simply sharing files across a group of people. In such a scenario, the cost of a server shutdown can be pretty high for an organization, as it would not only affect all the PCs connected to the server but also the productivity of the organization. Server manufacturers are coming out with various solutions to improve server reliability and to meet the increasing demand of high server availability across organizations. PCI hotplug is one step towards this.
So, what’s PCI hotplug, you may ask
PCI hotplug is an open, industry-standard technology pioneered by Compaq, and is increasingly being used in servers these days. It’s aimed at eliminating server downtime by making it easy to add, replace, and remove PCI cards while the server is up and running. Products with hotplug functionality primarily support three things: Hot replacement, Hot upgrade, and Hot expansion. Hot replacement is the process of replacing a failed or failing PCI card without shutting down the server. Hot upgrade is upgrading your PCI adapter card by using either an updated PCI card or by installing the latest updated driver for your PCI card. PCI hot expansion is nothing but adding additional PCI cards without shutting down the server.
PCI hotplug works best when used in conjunction with other fault-tolerant tools like RAID controllers. These tools keep the server running when a particular device fails by automatically transferring traffic to the standby device, and the hotplug feature lets you replace the faulty components without shutting down the machine. For example, if the primary network interface card of a server fails, then due to the fault-tolerant tools, the network traffic will automatically shift to the standby or secondary NIC (Network Interface Card) and the system will continue to run. Now, if the server supports hotplug functionality, it will notify the administrator of the failed NIC, which can then be replaced without shutting down the machine.
Thus, PCI hotplug improves productivity by giving you uninterrupted service. Plus, you save time in looking around for faulty devices in your server, making it fairly easily to manage and service various components.
Another advantage of PCI hotplug is that it’s compatible with previous PCI standards. So any new hotplug system hardware or adapter would also work with existing PCI-compliant systems. Also, since it’s an open industry-standard solution, multiple vendors can implement PCI hotplug solutions, giving you a much wider base of products to choose from.
What’s needed and how it works
To enable PCI hotplug functionality, you primarily need: a PCI hotplug-capable server, a hotplug-aware operating system (OS) and software, software drivers for each adapter card to support hot plugging, and a hotplug-aware user interface.
Both the hardware and software components work together to monitor the PCI system bus to identify failed PCI cards. They also make sure that any activity, such as adding, replacing, or upgrading a PCI card, takes place without affecting the rest of the PCI cards on the machine. Plus, they also protect the system from any adverse electrical effects.
When a PCI card fails, the system informs you with a system alert. You can then use the hotplug user interface, which is nothing but a software interface, to inform the OS of the faulty PCI card. The OS in turn instructs the hardware to isolate or power down that particular PCI slot from the rest of the PCI devices on that machine. (Instead of the software user interface, some machines also make use of a push button on the slots, to inform the software to initiate a power change).
Once the PCI slot is cut off from the rest of the devices, it can safely be replaced with a new one. Likewise, when a new card is inserted into the PCI slot, you can use the hotplug user interface to inform the OS of the new PCI card. The OS then instructs the hardware to power up the slot and connect it to the PCI bus again.
In addition, these systems are designed such that the addition and removal of adapter cards can take place easily. The slots have enough space between each other. Some machines also have handles on PCI slots for fast and easy addition and removal of PCI cards. In Compaq machines, for example, all slots are nicely arranged together with a convenient door, and two LEDs (Light Emitting Diodes) are placed above each PCI slot to identify their status. That is, a green light would indicate that the PCI slot is powered on and is functioning properly and an amber light would indicate that the slot requires attention. For further details on compatibility issues and system hardware, read the next article ‘PCI Hotplug in Action’.
Today, Microsoft, SCO, Novell, Compaq, and IBM are some of the leading vendors offering products based on this technology. Due to the increasing popularity and the benefits associated with this technology, you can expect more products from other leading vendors in the near future. Or, who knows, maybe even a superior, much-advanced technology.