Where should the Windows Manager spend the upgrade money?

We often face the following problem: Only x RMB can be used for Windows upgrades, But where should we use the money? What is the difference between the server and the desktop?

No matter when you allocate funds to upgrade the Windows system, it will be helpful to know where the funds will be spent and get the maximum return value. You should also pay attention to the time-changing priorities. It is not so useful to upgrade valuable to the present. Next, let's take a look at some common Windows upgrades and their respective values.

1. Increasing the memory will never hurt.

Increasing the memory in a Windows system will never produce errors, no matter how large or small the workload is. The larger the available memory, the more it makes up, such as slow disk access. This can be done by allowing more frequently-used data to be pre-cached.

This has become easier to understand in the past few years, especially after Microsoft's release of Windows Vista and Windows 7. These two operating systems are pre-filled with common data in RAM, and Windows XP has also done so to a certain extent. If an application requires additional memory, the data that is least frequently used in the cache is automatically migrated out.

As far as the desktop is concerned, the performance gains from increasing the memory will be balanced with the application load. 4 GB desktops that do not run many memory-intensive applications will not benefit from upgrading to 8 GB, such as Excel Data Tables and virtualization applications. This is also true on the server. It is easier to determine that the server needs to increase the memory, because its workload is naturally much larger than that on the desktop.

In terms of memory, your payment is rewarded, but this is only in some cases. Generally, it is not cost-effective to find the cheapest memory. That is to say, it is not worth spending on the most expensive memory. Only when the memory feature you purchased is required by the system, such as server memory correction, is the right time.

2. If you cannot increase the memory, upgrade the storage, but the buyer must pay attention to it.

Sometimes the system cannot handle the memory upgrade because it is overdrawn or because it is not cost-effective to increase the memory. This is a common problem in systems that have been used for several years. In this case, consider upgrading the system storage. This does not just mean adding more storage, but rather replacing a slow drive with a faster drive wherever possible.

In this environment, "faster" refers to the input/output IOPS per second ). You cannot take it for granted that this metric is directly related to the rotation speed or size. Remember, a larger drive may not be faster because of the seek time.

In addition, a faster drive speed is not necessarily faster than a large drive in the secondary partition. In short, you must use performance numbers to support storage instead of imagining what "faster" is.

It has become a trend to replace the traditional disk type with SSD. SSD does improve the original performance of servers and desktops, but there are still some warnings:

Cost. The price per GB for SSD is much higher than that for traditional disks. If you have 1 TB of system drives and use the same cost as a traditional disk, you may only get 1/5 or of the space on the SSD.

Controller bottleneck. Fast SSDS connected to traditional disk controllers, such as those regarded as standard SATA disks by the system, are basically limited by the throughput of the disk controller, rather than the disk hardware. High-end SSDS dedicated for servers have their own controller hardware as a solution to this problem. Warning in advance that you will pay thousands of dollars for such a device, instead of hundreds of dollars on a traditional SSD.

Read type. Compared with the turntable drive, SSD provides great improvements in the case of a large number of random reads. However, they do not have much improvement in reading the series. This is partly because of the controller bottleneck problem mentioned above.

3. Upgrading the processor is not worth a hassle.

In the past, the differences between different generations of processors were much greater than they are now, and they were easier to detect immediately. This time has passed, and most of the performance improvement has changed from processor to some functions. They have little to do with the speed of the original clock.

Virtualization is an increasingly important feature. The latest processor supports some features that enhance virtualization performance, such as the extension page table EPT ). These Processors may not be compatible with your existing system. If you are working on a rented machine group, you 'd better start from scratch rather than try the long-level existing hardware.

4. Full system upgrades should never be terminated, even if it looks expensive.

Check the amount of money you plan to use for the upgrade and try to calculate the total value. ROI may be more difficult to determine than you think, because it is difficult to objectively quantify the satisfaction of end users.

If you use a full system upgrade to achieve better results than an ordinary old-fashioned performance upgrade, we strongly recommend that you consider it. If you still have access to better and more urgently needed technologies, this is even more true.