Three methods for accelerating the WAN without the optimizer for internal data center Optimization

As IT continues to effectively centralize services, the dual-hot standby mode of data centers is becoming more and more common, and the network traffic of internal data centers is becoming more and more important. However, as backup traffic increases, virtual machine migration, service-oriented architecture, or replication between storage arrays, the business volume generated between data centers increases.

At the same time, the company's expectations for performance are also increasing: enterprises expect fast backup, synchronous replication and application response regardless of the situation between data centers.

When performance problems suddenly occur, internal data center communication becomes the culprit, and people must consider WAN optimization for these links. Optimization from an internal data center to a network is not the only option. IT staff should evaluate other schemes before deploying the WAN optimizer.

The following methods can improve high performance and WAN Link Utilization without buying internal data center optimization devices or software.

Deduplication technology. A particularly annoying performance issue is data replication between data centers and backup over the WAN link. In any situation, deduplication can greatly improve the performance by greatly reducing the volume of mobile data. The additional advantage is that deduplication can reduce storage consumption. Several data center managers interviewed by Nemertes all said that once de-duplication technology is well-established, they can avoid or eliminate the need for internal data center optimization.

Upgrade the protocol and use all functions. Even if the new version provides excellent performance, many enterprises still use the old version of core applications and WAN protocol. Sometimes, people use the Protocol in the simplest way and fail to take advantage of these features to reduce the traffic burden on internal data centers.

The best example is to upgrade the server information block protocol to server information block Protocol 2. Migration to the post-Vista version of the Microsoft file sharing protocol requires a lot of trouble. However, copying content through a high-latency link can greatly improve performance and reduce system overhead or additional capacity.

Other protocols can also reduce traffic, such as built-in compression functions. SOA Users communicate through Simple Object Access Protocol (SOAP) and sometimes generate multi-mb xml messages, which can be compressed by themselves. This reduces the data capacity by 90%.

When application software is developed within an enterprise, IT should make full use of network bandwidth through good application design. For example, you can accelerate the built-in SOA traffic: If you send data from service group A of Data Center 1 to Service Group B of Data Center 2, you can embed the data into the traffic that group A sends to Group B. However, if the data retrieved from the database can be replicated between data centers, you can extract group B data from the local copy so that the data is transmitted only once, instead of twice.

Obtain QoS. Finally, to improve the performance of the internal data center, the existing infrastructure can be implemented through the Qos function. Network staff can use some IP Baotou differential service code points (DSCP), with the server's network protocol stack, routers, switches, other network devices and service providers (if needed) tag data packets with specific traffic. Different levels of differential service code points can be used to indicate delivery priorities. For example, the differential service code point can ensure that storage replication does not need to wait when mail is transmitted between the mail server; it can also ensure that data packets are not discarded because resending will reduce application performance, this is very important for real-time protocols such as VoIP.

Finally, deploying WAN optimization devices or software is not the only way to optimize application delivery. By focusing on the infrastructure at other levels and improving performance, you can optimize WAN traffic between data centers.