Openstack: a solution to slow network access on control nodes.

Openstack: a solution to slow network access on control nodes.
Problem

In a dual-node environment, the instance access network on the "control node + computing node" is normal, however, instances on computing nodes copy files from other machines very slowly, with only dozens of K

Solution

Through the query of some information on the Internet, this may be caused by MTU. Therefore, MTU is reconfigured for all instances on the computer, and the result speed returns to normal, the instance uses the Windows2008 system. The configuration steps are as follows:

1. Start-run-enter regedit to open the registry. 2. in the registry, open the HKEY_LOCAL_MACHINE \ SYSTEM \ CurrentControlSet \ Services \ Tcpip \ Parameters \ Interfaces3. you will see several different sub-items in Interfaces, find the IP address of the sub-item, create a New DWORD Value, change the name to MTU search, right-click to modify and type 14544. restart an instance

Openstack Problems

OpenStack actually has three storage-related components. The degree of familiarity between these three components is the same as that of the component's appearance time. They are listed as follows:
Swift: Provides Object Storage, which is similar to Amazon S3 in concept. However, swift is highly scalable, redundant, and persistent and compatible with S3 APIs.
Glance-provides storage and management of Virtual Machine images, including many features similar to Amazon AMI catalog. (Glance background data is stored in Swift in the first practice ).
Cinder -- provides Block Storage, which is similar to Amazon's EBS Block Storage service. Currently, it is only used for Virtual Machine mounting.
(Amazon has always been an imaginary competitor and challenge object at the beginning of OpenStack design, so basically all key functional modules have corresponding projects. In addition to the three components mentioned above, for the important EC2 services in AWS, The OpenStack corresponds to Nova and maintains compatibility with EC2 APIs. There are different ways to achieve this)
Among the three components, Glance is mainly used to manage VM images. Therefore, it is relatively simple. Swift is very mature as an object storage service and supports CloudStack. Cinder is a relatively new block storage, with a good design philosophy and opportunities to combine with commercial storage, so the manufacturers are more active.
Swift
There are a lot of articles on the Internet except the official website about the architecture and deployment of Swift, which will not be repeated here. (you can also refer to the PPT that I previously gave a speech at the Shanghai site at the OpenStack China Line event ). From the perspective of development, there has not been much structural adjustment recently, so I 'd like to talk about the more suitable application fields.
According to the actual cases I have learned, there are four fields in Swift. (There should be more. I hope you can advise on the actual cases)
1. Network Disk.
Swift's symmetric distributed architecture and multi-proxy multi-node design make it a genetic fit for multi-user high concurrency Application Models. The most typical application is a network disk application similar to Dropbox, dropbox has already exceeded 0.1 billion users at the end of last year. For such access, a good architecture design is the root cause of support.
The Swift symmetric architecture makes the data nodes logically at the same level. Each node has both data and related metadata. In addition, the core data structure of metadata uses a hash ring. Consistent hash algorithms only need to relocate a small part of data in the ring space for node increase and decrease, which is highly fault tolerant and scalable. In addition, the data is stateless, and each data is fully stored on the disk. These factors guarantee the excellent scalability of the storage.
In addition, in combination with applications, Swift refers to the HTTP protocol language, which makes the interaction between applications and storage simple and does not need to consider the details of the underlying infrastructure, the application software can be extended to a very large extent without any modification.
2. IaaS public cloud
Swift's linear expansion in design, high concurrency and multi-tenant support make it ideal for IaaS, with a large public cloud, in this case, a large number of virtual machines are started concurrently. Therefore, for the background storage of Virtual Machine images, the actual challenge lies in the concurrent read performance of big data (larger than G, swift was initially used as the background storage of the Image Library in OpenStack. After years of practice in deploying thousands of machines in RACKSpace, Swift has proved to be a mature choice.
In addition, to provide the upper-layer SaaS Service Based on IaaS, multi-tenant is an inevitable problem. The Swift architecture design itself supports multi-tenant architecture, which makes it easier to connect.
3. Backup Archiving
RackSpace's main business is data backup and archiving, So Swift has been tested for a long time in this field, and they have extended a new business-hot archiving ". Due to the long tail effect, data may be called longer and longer time windows. Hot archiving can ensure that the Application archive data can be retrieved again in minutes, and several hours in the traditional tape drive archiving solution ...... remaining full text>
 
Help me summarize the problems in Enterprise Supply Chain Management

The Supply Chain Inventory Problem analysis facing the enterprise's external aspects are as follows:

First, the backward Supply Chain Inventory management cooperation model seriously affects the cooperation relationship and cooperation efficiency between suppliers and manufacturers. The inventory management cooperation in the supply chain is often subject to the short-term interests of strong enterprises. Most supply chains still adopt the form of inventory management by users themselves, suppliers only obtain rough monthly demand forecasts, Temporary urgent orders, and monthly consumption and inventory check information through outsourced personnel. The two parties did not conduct research on the joint replenishment strategy, there is no need for information sharing. As a result, the scalping effect is significant, greatly increasing the inventory level of the entire supply chain.

Second, members of the supply chain are very prominent. Including downstream nodes squeezing upstream nodes, strong nodes squeezing weak nodes and similar nodes squeezing each other. This squeeze is represented by price, cost, and risk squeeze, which eventually forms the squeeze of profit and living space. This squeeze is not only a blow to the weak enterprises, but also has a negative impact on the supply chain. It usually increases the overall cost of the supply chain and affects the competitiveness of the supply chain.

Third, there is a serious crisis of trust among supply chain members. This includes downstream nodes that do not trust upstream nodes, asymmetric trust desires of upstream and downstream nodes, the destructiveness and vicious circle of trust, and the lack of a standardized trust mechanism in the supply chain. In the Supply Chain Inventory control operation, there is no standardized mechanism to constrain the standards that enterprises in the supply chain should provide their own information for the supply chain, it also assists neighboring enterprises in achieving efficient and smooth logistics connection.

Insufficient utilization of external logistics resources, and poor logistics capabilities seriously affect enterprises' inventory control capabilities. At present, many enterprises have foreign procurement, but the long lead time of procurement is a major problem for most enterprises. If you make full use of external logistics resources, you can greatly shorten the lead time of procurement.

Analysis of supply chain inventory problems for Enterprises

The internal problems facing enterprises are mainly discussed in three aspects: first, the internal logistics decision-making mechanism of enterprises violates the basic idea of inventory control; second, the internal performance evaluation and Incentive Mechanism of the enterprise impede the rationalization of the logistics operation and control of the enterprise; third, the technical conditions of the enterprise's internal inventory control are quite backward with the environment.

Research found that, the root cause of the impact on enterprise inventory control is that the enterprise's behavior in the logistics decision-making process, such as the production plan, demand plan, and procurement plan, violates the inventory (or overall logistics) of the enterprise) the idea of control. Figure 1 is the decision-making process of Case 1. At present, many manufacturing enterprises playing the supplier role in China adopt this production/inventory management decision-making method. The logistics of enterprises is mainly driven by manufacturers' monthly orders, and the necessary management and control of inventory to respond to the manufacturer's timely needs for emergency replenishment. Such enterprises often have a high average inventory level and frequent emergency replenishment, making it difficult for them to shoulder and cope with such problems. Figure 2 is the decision-making process of Case 2, which is adopted by many end product manufacturers. Such enterprises often set production targets, set rolling demands for production, and set material procurement based on rolling demands and supply market prices. Finished Products depend on the sales performance. In this way, the warehouse is passively stored in and out of the warehouse, and there is no inventory control in essence.

The impact of internal performance evaluation and incentive mechanisms on Enterprise Inventory Control is relatively concealed, but enough attention should be paid to it. A typical phenomenon in enterprises can fully illustrate this problem. For example, the production department determines the excess production in the original plan and submits materials to the raw material warehouse because the original supply plan does not have this part of the supply plan. On the surface, raw materials are out of stock. The warehousing Department issued a notice to the supply department for the Supply Department to organize procurement. The Supply Department said that the procurement cycle was nearly two months, and it was very difficult to do a good job in this supply. Therefore, the Supply Department applies to allow the purchase of large quantities of materials for storage. As a result, the inventory level of the enterprise increases naturally, the logistics turnover slows down, and the inventory and related costs increase. Analyze the cause of this phenomenon and find that there is such a causal relationship, because there is a bonus in the super production, at least there is no punishment, so the production department will appear super production behavior. On the surface, however, the supply department is subject to certain penalties during the assessment. For the sake of its own interests, the Supply Department always tries to increase the procurement volume within the scope permitted by the procurement funds, deliver the information to the warehousing department and give a lot of room when reporting the procurement lead time, so that you can get a higher assessment or less penalty during the assessment. The warehousing Department falls into the warehouse due to objective reasons such as location... the remaining full text>
Liu qingcloud [authoritative expert]