FREE BOOK

Chapter 8: Data Storage Design and SQL Server

Posted by Apress Free Book | ADO.NET & Database December 16, 2008
This chapter will explore the most common storage technologies available to you. It will also present some benefits and concerns about different storage designs so that you can more easily choose what suits your solution. We will finish the chapter with a presentation of SQL Server-examining its architecture and how it fits into your storage policy.

A Middle-Way Design
 
The last example is a middle-way data storage design. Here you use various technologies to store your data. Some of it is stored on DAS, some on SAN, and the rest on NAS appliances. This solution is very flexible, and offers several possibilities for your projects. You can let team leads on each project decide what best serves their purposes. 
 
This could be a more complex design to implement, however. Many design considerations have to be made, just as with the distributed model, making the process long and complicated before it can be implemented. Without proper guidelines, it will also be costly. Since this solution involves a lot of storage technologies from different vendors, TCO risks being higher than a centralized solution.
 
Choosing Your Storage Solution
 

Once you have determined the requirements for your storage solution, you must decide how you want to go about meeting those requirements.
 
Reasons for Choosing a DAS
 

A DAS is a popular choice for storing data. It is a solution with a low entry cost, because hard drives are cheap these days. It is tempting for a department or a workgroup to choose this technique, since it is easy to implement. A small organization or department of a larger organization often does not want to spend time and money on shared storage or on the enhanced availability and higher performance that comes with it. Our view is that this might harm the organization in the end. A large number of DAS solutions might very well be complex and expensive to manage. The hardware often comes from different vendors, so standardization might suffer. This could lead to isolated data islands that might not be accessible to more than the server where the hardware is attached. From an enterprise view, such a solution is not what you should aim for.
 
On the other hand, a DAS solution is easy to launch. Most administrators feel at home with this kind of technology. Another issue is performance. Because DAS is directly attached to the server, performance is great. Depending on the data, it can also be hard to get more than one server to access it at the same time, so a SAN or NAS appliance might be a better choice if you need the data to be available for many servers or applications. And in a large enterprise, you can bet that this is the case. It is true, however, that a database with database files stored on a DAS can be accessed by many applications, but then you will have the resultant manageability issues to deal with. You must be aware that access to the data is limited to the number of physical network connections the server can handle at the same time, however. 
 
We would recommend using DAS when you need to have a database close to your Web server, and for use as a caching server-that is, a server that you only use for data immediately needed for your Web solution, thereby providing fast access to this data. All your user data, customer data, product data, and so on should be stored on a more permanent basis in a centralized data store like a SAN. The contents of your user sessions information, shopping baskets, and the like should be stored in your DAS caching database. This way you provide fast access of such info for your Web solution.
 
Reasons for Choosing a NAS
 

Network-attached storage is a scalable and flexible way to fulfill your enterprise file-sharing needs. This technology, like DAS, offers great simplicity, and furthermore most administrators know about it. You can easily administer security and access control for this solution. Access to a network-attached storage system is limited to the LAN, however, and if you do not want to use Gigabit Ethernet, this can be a potential bottleneck. 
 
Compared to a SAN, a NAS system is simpler and less expensive. SANs require additional hardware and cables, and a lot more effort has to go into the design of it. The fact that you attach it directly to the LAN infrastructure also makes it complex to implement.
 
In addition, you might run into problems with your database applications, because not all of them support storing database or log files on network-attached storage appliances. 
 
Your backup solution is something you also have to consider. It must allow you to back up over the LAN; otherwise, you might run into trouble. Even if it allows you to back up over a LAN connection, you must plan your backup schedule carefully. You do not want to run your backups when network traffic mostly consists of business-critical operations.
 
Reasons for Choosing a SAN
 

SANs make it possible to implement highly available storage subsystems to multiple hosts. SAN fabrics are networks that connect hosts to storage devices. By using fabric switches, you can connect separate SAN fabrics, or SAN islands as they are sometimes called, to each other. 
 
One of the benefits of SAN is that you can expand it quite easily. You can add new storage devices to your SAN when you need to without interrupting the existing structure in the meantime. This means you can scale the SAN when the need for it arises.
 
Another benefit is that backup and restore operations are done over the SAN fabric. This way you get high speed on these actions and at the same time minimize impact on the LAN. If you want to implement a server-free backup, you can do so by using the appropriate hardware, but you should be aware that this takes a few extra configurations to maintain integrity of your application data. 
 
It is also possible to manage the SAN from one place, which makes manageability easier. You can pool storage together and allocate it to those servers that need it. 
 
One of the problems with a distributed model has been that it often has a lot of excess storage, and thus does not utilize resources as well as it could. With a SAN, the excess storage can be fully used, and nothing is wasted. 
 
All this might sound great, but there are, of course, disadvantages. A SAN is very expensive to implement, as you have seen. Luckily, most of the expense occurs at the initial implementation, and the centralized management as well as the reduction of wasted storage and greater flexibility makes the TCO decrease over time.
 
NOTE One of our customers has implemented a SAN over the last year. Although it has cost a lot of money, it seems like administrators and users alike are satisfied with the solution. At the same time, the customer has implemented SharePoint Portal Server and now has a truly centralized solution in place.
 
A SAN also demands a lot more from your administrators, since they might need to learn a new technology. If a SAN is not already implemented in the company, once you start planning for it, you soon find that it is a complex solution. A SAN requires specialized knowledge to design, administer, and maintain. 
 
If you could choose, a SAN would be perfect for your applications, but unfortunately the choice is not always yours to make. Implementing a SAN requires an enterprise to develop a new storage architecture and policy. Since the initial costs are high, many might hesitate at this, especially since the long-term perspective often is forgotten in the cost-aware times in which we are living.
 
Consider Combinations
 

Rather than choosing just one the preceding technologies, you could use various combinations of these technologies. Many new NAS devices also handle access to SANs (see Figure 8-4). This increases their scalability, while still offering the benefits of network-attached storage. You can also improve backups and restores, because this setup lets you perform these operations to backup devices on the SAN. This way, you get server-free backup of your network-attached storage devices as well.
 
 
 
  Figure 8-4. A NAS appliance that also provides access to SANs 
 

By using this combination design, you can take advantage of the best features from the different storage technologies. At the same time, you get a great deal of flexibility and high performance. This combination is not perfect, however. The price tag is not something everybody could afford. You must also be aware of the increasing complexity in your solution, since you mix hardware from different vendors. Naturally, this affects operability and manageability. 
 
You have to consider what kind of storage solution you actually need and the requirements your business has of your storage before you run out and purchase anything. Design is always something you should not take lightly, and it is often worthwhile to take a little extra time to think about it early on. 
 
Next we will move on to our discussion of SQL Server. SQL Server is a powerful database manager, and it is important to know how it works and how it can be used in a storage policy. We will show you the different versions you can use, examine the architecture of SQL Server, and share some tips for working with SQL Server that we have learned over the years.

Total Pages : 12 12345

comments