Theoretical Foundation 2.1 Data Backup

Transcript

THEORETICAL FOUNDATION 2.1 Data Backup In general, backup in computing system can be defined as the process to help or support the active main system in case of it occurring a failure to function. More specifically, the data backup can be considered as the activity of copying files or databases to the backup media in order to preserve it preventing occurred data loss in the main system [5]. The backed-up data can be reside in the backup media in the same form or achieved depends on the need. Once the data backed-up, there must be a way to retrieve it from the backup media to the main system or other. The process can be said as backup restoration or recovery. Backup is usually a routine part of the operation of large businesses with mainframes as well as the administrators of smaller business computers. For personal computer users, backup is also necessary but often neglected. A survey from Symantec in 2009, regarding to information protection in Small Medium Business, shows that annually those companies spent up to $16,000 for backup, recovery and archival, and $10,000 for disaster preparedness [2]. There are many methods to perform data backup which will impact the performance of the backup process including the speed of the data transfer, the comprehensiveness of the backed-up data, the storage needed, the easiness of data restoration, and the network performance (if any). At last, those factors will also determined the cost that the company have to spent. 19 2.2 Classification of Data Backup Method To classify the data backup approach, we need to determine the base on how the data backup can be done. The author synthesize the categories into 6 classification of methods based on their architectures, storage locations, medias, frequencies, scales, and attributes. The following figure 3 will maps the synthesize classification of data backup method. Figure 3: Classification of data backup 20 2.2.1 Backup Architecture Regarding to its architecture, backup can be classified as directly attached backup, client-server backup, and storage area network backup. Table 1 shows the brief comparison about the backup architecture characteristics. Directly Attached Client-Server SAN Transfer rate Fast Slow Fast Implementation Easy Moderate Difficult Management Difficult Easy Easy Cost Low Moderate High Table 1: Comparison of backup architecture 2.2.1.1 Directly Attached Backup As its name, this approach is done without using network connection. In other words, the administrator need to directly attach the backup media into the backup client as the secondary storage [6]. The media can be in form of external hard disk, tape, DVD, etc. Generally, the directly attached media should have the fastest performance during the data transfer, compared to the network media such as gigabit ethernet or fiber optic cable. Despite of the considerably fast transfer rate, non-network approach might become troublesome when the administrator have to handle many backup clients. The one who handle the backup have to attach the media one to another backup client. Without aided software, it is difficult for the admin to manage such backup rotation that could make the backup management to be disorganized. This non-network approach can offer a low cost solution needed to just 21 provide the backup media. Due to its low difficulty to implement, this approach might use by entry level enterprise that does not implement any backup management before. There also a probability that large enterprise done nonnetwork backup to complement the more advance backup solution. 2.2.1.2 Client-Server Backup Figure 4: Client-server backup based on Preston The client-server backup consist of backup clients, which have data to be backed up, and backup servers, which performing the backup and hold the data. It is often for a company to have many to one relationship in this kind of architecture. In other words, they can specify a dedicated server to backup some other servers or workstations shown in figure 4. Those clients and server are connected through a network connection, can be in various network type. The backup server provide a flexibility in managing the backup. There are options to choose such enterprise level operating system to cope the environment. There are also numerous of software with various backup algorithm to aid the admin. Having a centralized control also provide easiness to handle the clients 22 scalability. A server also can be installed with many options of backup media such as SATA, SAS hard disk, or using tape drive to keep the backup image or a plain form of backup. The drawback of this kind of method is that the client and server are mostly connected to TCP/IP network. It cause a bottleneck in term of transfer rate from a quick transfer input such as SAS into the slower output such as gigabit ethernet. This client-server approach can offer a middle cost solution needed to provide the backup server and its storage. Due to its moderate difficulty to implement, this approach might use by middle to high level enterprise that might run several servers and many workstations during its daily operations. 2.2.1.3 Storage Area Network Backup Figure 5: SAN architecture based on Preston The SAN backup typically require a set of storage media, might be the combination of disk drive and tape drive to form a certain tape or disk library. SAN commonly managed by a special software, speaks its own protocols, and connected through the backup client with such SCSI and fiber channel shown in 23 figure 5 [8]. This feature make the mechanism in the data transfer as if the storage is connected internally within the client. While not just mimic the transfer mechanism, it actually highly affecting the transfer rate of the data. Unlike the client-server approach that generally speaks TCP/IP and uses media like ethernet, this SAN uses protocol such as SCSI3 that enable the the disk on the client speaks directly with the backup tape/disk library. This feature eliminates the bottleneck disadvantage of the client-server approach. This SAN approach require higher cost solution needed to provide the SAN system itself. Due to its higher difficulty to implement, this approach might use by higher level enterprise that runs several servers with large amount of data to backup. 2.2.2 Backup Storage Location Regarding to its storage location, backup can be classified as local area network backup, wide area network backup, and cloud backup. Table 2 shows the brief comparison about the backup storage location characteristics. LAN WAN Cloud Transfer rate Fast Moderate Slow Implementation Easy/Moderate Difficult Easy Management Moderate/Difficult Moderate Easy Cost Moderate High Moderate Disaster Recovery No Yes Yes Full Control Yes No Yes Table 2: Comparison of backup storage location 24 2.2.2.1 LAN Backup A backup system can be said doing this approach where the backup media and the clients located in the same area of building, or connected within a local area network, or can be said as one network. The network media can be in form of directly attached storage or LAN media such as fast ethernet, gigabit ethernet, or fiber optic cable. This approach should have the fastest performance compared to the other storage location since distance is heavily influence the data transmission. Having a close location to its backup media make this solution provide responsiveness in handling backup clients within a local network. The one who handle the backup have to manage the backup media, probably a server and clients through the network. There are softwares to aid the admin to manage such backup rotation that could make the backup management more organized. LAN backup approach can offer a middle cost solution needed not to just provide the backup media, but also to have more proper network implementation and management such as the bandwidth management [9]. This approach might suitable by entry to middle level enterprise that do not have other branches in other area. 2.2.2.2 WAN Backup A backup system can be said doing this approach where the backup media and the clients located in different area of building, but still belong to the company, connected across the network, or can be said different network. The technologies can be in form of internet or other type of WAN such as VPN [10]. Practically the transfer rate can be slower compared to the LAN backup. There are also security 25 issues faced if the data transfer across the internet. In term of backup management, WAN backup achieve the same advantage as the LAN backup, and the software used will be much more the same with the LAN backup, probably with different configuration. Plus, it is good to implement such centralized backup management where the backup clients located across different area. Since the backup media located in the different area, WAN backup also provide disaster recovery in order to keep data available if the main system struck by disaster. WAN backup approach could raise higher cost solution in addition to provide the a sufficient WAN technologies considering its bandwidth, security, etc. This approach might use by higher level enterprise that might runs several distributed servers for its daily operation. 2.2.2.3 Cloud Backup Cloud backup is basically a WAN backup which backup media belongs to a backup solution provider. The technologies will be much more the same like WAN backup, with high probability of using internet. The transfer rate is highly dependent on the internet traffic. There are also facing security issues the same as WAN backup, moreover the enterprise itself can not have full control of the media. In term of backup optimization, cloud backup could achieve slightly less advantage as the WAN backup due to partial control of the backup media. But from the efficiency, cloud backup offers an easiness since the company will not have to maintain the backup system [11]. Cloud backup also provide disaster recovery just like WAN backup. 26 Cloud backup can offers cheaper solution than WAN backup, usually the backup solution company charge based on the data size that backed up. This approach suitable for low to middle range company which might not have immense data and allowed to use 3rd party service for data storage. There are some reputable cloud backup provider such as Ahsay, Asigra and Zmanda. 2.2.3 Backup Media Regarding to its storage media, backup can be classified as tape based backup and disk to disk. Table 3 shows the brief comparison about the backup media characteristics. Tape D2D Transfer rate Moderate Fast Implementation Moderate Easy Management Moderate Easy Cost Low Moderate Durability High Moderate Table 3: Comparison of backup media 2.2.3.1 Tape Backup The tape backup is a form of data backup that is used to create a image of the data stored in a system at a specific point in time. The data are copied onto a reel of magnetic tape and can be used for archival purpose for future references. Tape based backup often use a rotation strategy which enable the backup system to have several snapshots at a time of current main system data [12]. There are several types of tape drive, for instance is LTO with the transfer rate ranged from 30 MB/s to 280 MB/s [13]. Most tape backup systems allow for 27 sequential access only. The data are accessed at the beginning of the captured data and then moved through in the order it was recorded. Tape drive also can be more durable and shock resistant compared to disk drive due to insensitive tape's magnetic band and its sequential mechanism. It is not unusual for many enterprises to perform a tape backup since it consume low cost while also performing the other type of backup. For example, a company might also performing WAN backup to another server, providing quick data restoration or fault tolerance when the main system fails. A strategy of this type results in a great deal of data security for the company, making a slight possibility for data to be permanently lost. 2.2.3.2 Disk to Disk Backup The D2D backup is an activity of data backup that is using a hard disk drive media, can be in form of an image or a plain folder of files. The data are copied into a magnetic disk and can be used for quick responded backup data in case of data loss occurred in the main system. D2D backup might not use rotation strategy and just overwrite or append the existing data content of the main system. There are several types of disk drive, for instance is SATA2 with the average interface speed of 3 Gbit/s and SAS with average 6 Gbit/s [14]. Most D2D backup systems allow for random access data. Disk drive might be less durable and less shock resistant compared to the tape drive due to its random access mechanism that causes the disk to keep rotating in high speed. Common failure of disk drive includes bad sector and another type of crashing. D2D backup cost higher than the tape backup. Enterprises utilize D2D 28 backup for various strategies. For example, a company might also performing a plain copy of directories as well as performing WAN backup to another server, providing quick data restoration if there is a minor data loss occurred. A strategy of this type provide more redundancy of the data resulting a flexibility in the backup mechanism. 2.2.4 Backup Frequency Regarding to its frequency, backup can be classified as hot backup, warm backup, and cold backup. Table 4 shows the brief comparison about the backup frequency characteristics. Hot Warm Cold Availability High Moderate Low Consistency High Moderate Low Network Requirement High performance Moderate performance Low / no performance Implementation Difficult Moderate Easy Cost High Moderate Low Table 4: Comparison of backup frequency 2.2.4.1 Hot Backup Hot backup requires the backup system up for 24/7. There is no time needed to trigger the system to allow it performing the backup process since it should be done automatically [15]. The idea of this kind of backup is to perform backup process in real time whenever any data is changed in the main system. Although it is said to perform real time backup, in practical there are no such thing 29 which is real time due to the latency of the network media and so on. The need of performing backup in real time make this method requires a considerably high network performance. It might be no problem if the backup media located locally which have low latency in transferring data, otherwise the implementor should provide high bandwidth and low latency network, probably a dedicated one to make sure the reliable data transfer. Hot backup is commonly used by banking and commerce company where the large portion of the data is belong to the customers and highly valuable to be recorded. For instance, a commerce company might uses an database application that record its customer transactions. It performs the MYSQL hot backup to perform a replication every time there is a new transaction occurred. Having those features of highly available backup system, hot backup have to be paid off with rather high cost. It might be needed to provide network with high performance, and such backup media that have reliable performance in 24/7. 2.2.4.2 Warm Backup Warm backup might also requires the backup system for 24/7, or in less frequent range of time. The system might required some time, or no time to trigger the backup process, depends on the availability of the system [15]. The idea of this kind of backup is to perform backup process in a non-outdated range of time and might use auto scheduling, such as daily backup and weekly backup. As a result, the degree of consistency in the backup media is not as comprehensive compared to the hot backup. Since there is no need of performing backup in real time, this method can 30 compromise the need of high network performance. Same case as hot backup, it should be no problem if the media located locally. When it comes to a remotely located backup media, the latency can be compromise, but at least it should provide adequate bandwidth to handle large scale of backup before the next backup schedule due. Warm backup is more generally done by middle-up level enterprise where its data mostly still belong to the internal like documents or assets database. For instance, a company might uses daily auto scheduling to backup folders of documents, and weekly backup for the its asset database. The backup process might be executed after work hours in order to have significant changes, and avoid high network traffic during the work hours. This kind of backup might offer midline cost, lower than the hot backup, due to the less need of network performance and less reliable backup media that not working for 24/7. 2.2.4.3 Cold Backup Cold backup is the least frequent method to conduct backup. The system does not have to be up regularly or for certain range of time. It may be up only when the backup process needed. It might take longer time to trigger the backup process. [15] The idea of this kind of backup is to perform backup process once in a while, as it needed. The purpose might be various, including for archival purpose. High degree of consistency in the backup media is not a concern for cold backup. From its purpose, this method might not need network availability since it 31 only performs backup once a while, and it often use the directly attached backup media. The concern here is generally focused on the capacity and scalability of the backup media, whether it can fit the data or not. Cold backup more widely used by small office-home office (SOHO), or by personal use which the growth of data is not immense exponentially. Some middle level enterprise also done it due to the unawareness of data loss or insufficient resource. Higher level enterprise also might done this approach for archival purpose. This kind of backup offers lower starting cost that used for backup media procurement. There is no need for installation of network media or more advanced backup media. 2.2.5 Backup Scale Regarding to its scale, backup can be classified as full backup, differential backup, incremental backup, and mirror backup. Table 5 shows the brief comparison about the backup scale characteristics. Full Differential Incremental Mirror Comprehensiveness High Moderate Low Most Backup Speed Slow Moderate Fast Slow Restore Speed Fast Moderate Slow Fast Storage Required Large Moderate Low Large Table 5: Comparison of backup scale 32 2.2.5.1 Full Backup Full backup performs the backup process by copying the entire content of specified filesystem or directories. The backup content considered as comprehensive and self-contained since it replicating all specified files every time the backup process triggered. It consumes large amount of space in the backup media since it copy all specified files, plus the number of rotations used the system. As a result, the backup size can be multiply compared to the the original size of the specified data. The example of having full backup alone can be explained as following. Suppose we perform full backup to a folder with 500 MB size. At the second day, it grows 200 MB. When we perform full backup for the second time, it actually copy as much as 700 MB, and so on if the data keep growing. 1200 1000 800 600 Backed up 400 200 0 1st day 2nd day 3rd day 4th day Figure 6: Full backup Since this method always perform entire copying, it will take longer time to backup than differential and incremental backup. The backup result will always be the most comprehensive in the latest rotation. As a result, the restoration 33 process will become easier and faster than the differential and incremental because the administrator only have to choose the latest full backup. Full backup alone can be comfortably used when the data is relatively small and not have immense growth rate. 2.2.5.2 Differential Backup Differential backup performs the backup process by copying all files changed since the last full backup. The backup content is complementing the full backup since it only replicating all changed files every time the differential backup triggered [16]. It might consumes less amount of space in the backup media depends on how much change occurred since the last full backup. As a result, the backup size can be reduced compared to the full backup. The example of having differential backup can be explained as following. Suppose we perform full backup to a folder with 500 MB size. At the second day, it grows 200 MB. When we perform differential backup at the second day, it actually copy as much as 200 MB. If the data grows another 300 MB on the next day, than the differential will perform the copy of 500 MB, and so on. 34 900 800 700 600 500 Backed up 400 300 200 100 0 1st day 2nd day 3rd day 4th day Figure 7: Differential backup The feature of copying only the updated data makes differential backup need less time to complete compared than the full backup if the updated data is not immense. The backup result is only comprehend the last period of backup. As a result, the restoration process of a comprehensive data will become slower, because other than the need to specify the latest one, it also need the result of full backup to complete the data. The combination of full and differential backup can be comfortably used when the main data is large and not have immense growth rate. 2.2.5.3 Incremental Backup Incremental backup performs the backup process by copying all files changed since the last full backup or differential backup. The difference here is that the incremental backup works with level that can limit the growth of data per level [16]. When the growth of data exceed the limit, it will increment the backup level and reset the limit. The backup content also complementing the full backup. 35 It might consumes less amount of space in the backup media depends on how much change occurred since the last full backup. As a result, the backup size can be reduced compared to the full and differential backup. The example of having incremental backup can be explained as following. Suppose we perform full backup to a folder with 500 MB size and have limit the growth rate at 100 MB per level. At the second day, it grows 200 MB. When we perform incremental backup at the second day, it actually copy as much as 200 MB to level 1. If the data grows another 300 MB on the next day, than the incremental will perform the copy of 300 MB to level 2, and so on. 600 500 400 300 Backed up 200 100 0 1st day 2nd day 3rd day 4th day Figure 8: Incremental backup The feature of limiting the growth rate in each level makes incremental backup need less time to complete compared than the differential backup if the updated data is immense. The restoration process of a comprehensive data will become slower and more troublesome, because other than the need of last full backup, it also need the result of the latest backup on each level to complete the 36 data. The combination of full and incremental backup can be comfortably used when the main data is large and have immense growth rate. 2.2.5.4 Mirror Backup Mirror backup is similar to the full backup in terms of operation, it also performs the backup process by copying the entire content of specified filesystem or directories. The difference is in the result of backup that will be in form of plain file, not compressed or in other special format. It consumes large amount of space in the backup media exactly the same as the one in the main system. This kind backup content considered as the most comprehensive and selfcontained since it also replicating the files attribute such as read, write, execute permission, etc. That feature could be useful for the system which requires quick restoration of the data. 2.2.6 Backup Attributes Regarding to its attributes, backup can be classified as compressed backup and encrypted backup. Table 6 shows the brief comparison about the backup attributes characteristics. Compressed Features Technology Encrypted Reduce storage size Provide confidentiality Server/Client side Server/Client side .tar .zip .gzip .iso TLS, SSL, GPG Table 6: Comparison of backup attribute 37 2.2.6.1 Compressed Backup Compressed backup is the process attempting to reduce the size of backed up data than its original data or folders in the main system. The compression can occurred in the client side or in the server / backup media side. The quality of the compression usually affected by the CPU or processing overhead in the machine. It is a tradeoff, the more it get compressed, the more CPU power and time needed to done the process. The client side compression is useful when the client is not heavily loaded with other task, resulting the transmitted data to be reduced, could reduce the network traffic used to backing up, and finally speed up the backup process. While the server side compression can be used in the opposite circumstances. The format of compressed backup would be various depends on the used software, but many software cope it to known compression format such as .tar, .zip, .bzip, .gzip, .iso, to provide easiness during the restoration. 2.2.6.2 Encrypted Backup Encrypted backup is the process attempting to convert the content into some code known as cipher in order to provide the confidentiality for the backed up data. This method suitable to perform when the company conduct a WAN or cloud backup with no secure network or private network. There are two things to consider, whether performing the encryption during the data transmission only, or to perform encryption of the actual contents even after the data already stored in the backup media. 38 If we have full control of the remote backup media, the option of encrypting during the transmission would be suitable. This can be achieved by sending the data through a SSL tunnel. The advantage of this method is during the restoration, the key to do decryption process is no longer needed. However, if the company does not have control over the remote backup media, the other option would be suitable. One option to perform this is by using symmetric encryption either on the client or server side. As a result, the admin need to either move the key, or the backed up data to the respected machine. 39 2.3 Example of Enterprise Backup Software Based on the author research about renown enterprise level backup software, there are 3 softwares that comes into the comparison. Table 7 shows brief comparison of Bacula, Amanda, and IBM Total Storage Manager which is partly taken from Bacula's community research [17]. Bacula Amanda Yes Client-Server, SAN** IBM TSM No Client-Server, SAN** LAN, WAN LAN, WAN, Cloud LAN, WAN Tape, Disk, DVD Tape, Disk, DVD Tape, Disk Hot*, Warm, Cold Hot**, Warm, Cold Hot, Warm, Cold Backup Scale Full, Differential, Incremental, Consolidation Full, IncrementalDifferential Full***, Incremental Backup Attribute Compressed, Encrypted (TLS) Compressed, Encrypted (SSL, AES, PGP) Compressed GUI Yes - bat Yes - ZMC** Client & admin client Yes Yes Yes Yes Yes Yes** No* Yes** Yes** Open Source Backup Architecture Backup Location Backup Media Backup Frequency Multi Platform MS Exchange Support MSSQL & Oracle Yes Client-Server Table 7: Comparison of enterprise backup software *) No built in feature, Bacula community provides the script to enable the feature **) Need to purchase additional license ***) Full backup only done at the very first time backup 40 2.3.1 Bacula Bacula [18] is a powerful Linux backup solution, and it’s one of the few Linux open source backup solutions that’s truly enterprise ready. But with this enterprise readiness comes a level of complexity you might not find in any other solution. Unlike many other solutions, Bacula contains a number of components:  Director. This is the application that supervises all of Bacula.  Console. This is how you communicate with the Bacula Director.  File. This is the application that’s installed on the machine to be backed up.  Storage. This application performs the reading and writing to your storage space.  Catalog. This application is responsible for the databases used.  Monitor. This application allows the administer to keep track of the status of the various Bacula tools. Bacula is not the easiest backup solution to configure and use. It is, however, one of the most powerful. So if you are looking for power and aren’t concerned about putting in the time to get up to speed with the configuration, Bacula is your solution. 2.3.2 Amanda Amanda [7] allows an administrator to set up a single backup server and back up multiple hosts to it. It’s robust, reliable, and flexible. Amanda uses native Linux dump and/or tar to facilitate the backup process. One nice feature is that Amanda can use Samba to back up Windows clients to the same Amanda server. It’s important to note that with Amanda, there are separate applications for server and client. The key 41 components of amanda includes like following.  Amanda index server, which performs the backup process by sending dumpers to the backup clients  Holding disk, which holds the backed-up data before it is flushed to the tape device, enabling the server in executing backup from many clients at a time.  Virtual tape device, which enables the hard disk drive used for backup to mimic the tape backup rotations.  Tape type and dump type that can be specified flexibly with many options such as the maximum storage, the network usage, the compression and encryption type. 2.3.3 Tivoli Storage Manager IBM Tivoli® Storage Manager [19] provides a wide range of storage management capabilities from a single point of control, helping companies ride the information tidal wave which have following key features.  Backup and recovery management, which helps the admin to perform intelligent backups and restores utilizing a revolutionary progressive incremental backup and restore strategy, where only new and used files are backed up.  Hierarchical storage management, which helps the admin to perform policybased management of file backup and archiving.  Archive management, which helps the admin to easily protect and manage documents that need to be kept for a certain period of time. 42  Advance data reduction, which Combines progressive-incremental backup, source and target data de-duplication, compression and tape management to provide best-in-class data reduction 2.4 Example of Backup Media 2.4.1 LTO Ultrium Tape This Linear Tape Open technology was developed jointly by HP, IBM and Certance (Seagate) now Quantum to provide a clear and viable choice in an increasingly complex array of tape storage options. Ultrium LTO technology is an "open format" technology, which means that users will have multiple sources of product and media. The "open" nature of LTO technology also provides a means of enabling compatibility between different vendors' offerings. Generally, LTO cartridges able to do one of these two mode of operations, either it is rewriteable (RW) or write once read many (WORM). Other than that, LTO 5 also equipped with native 2:1 compression and 256 bit AES encryption [20]. Table 8 provides the example of the latest generation LTO which is LTO 5 Ultrium. HP IBM Quantum Capacity (Compressed) 3 TB 3 TB 3 TB Transfer Rate 280 MB/s 170 MB/s 280 MB/s Mode RW, WORM WORM WORM Table 8: Example of LTO 5 Ultrium specifications 43 2.4.2 External Hard Disk There are desktop external hard drives are based on the 3.5-inch internal hard drives and laptop (or portable) external hard drives that are based on the 2.5-inch internal hard drives. Generally, external hard drives are connected to a computer using collectively these types of connections: USB 3.0, USB 2.0, FireWire 400, FireWire 800, and eSATA [21]. Portable external hard drives are also often bus-powered, meaning it require only one cable for both data and power connections. Table 9 provides example of current generation of external hard disk. WD MyBook 3.0 Capacity 1 TB Iomega eGo Portable Seagate FreeAgent GoFlex Pro Transcend StoreJet 25 Mobile 500 GB 500 GB 500 GB Interface type USB 3.0 USB 2.0, FireWire, FireWire 800 USB 2.0 USB 2.0 Other features Security lock slot Drop Guard N/A One Click Backup, Shock Resistant Table 9: Example of external hard disk specifications 2.4.3 Server Hard Disk Server hard disk generally divided by 2 categories according to its interface which is SATA (Serial Advanced Technology Attachment) and SAS (Serial Attached SCSI). SAS hard disk is suitable for critical operation with high availability that requires immense I/O, related with backup, it more likely suitable with the hot backup operations. While SATA would be sufficient for less critical operation and hot or warm backup with less I/O. Table 10 provides example of these server hard disk. 44 HP ProLiant SATA HP ProLiant SAS IBM xSeries SATA IBM xSeries SAS Capacity range 120 GB – 2 TB 72 GB – 600 GB 160 GB – 500 GB 73 GB – 260 GB RPM range 5.4 K – 7.2 K 7.2 K – 15 K 7.2 K 10 K – 15 K Transfer rate 1.5 Gb/s – 3 Gb/s 3 Gb/s – 6 Gb/s 2.8 Gb/s 6 Gb/s Table 10: Example of server hard disk specifications HP : http://h18004.www1.hp.com/products/servers/proliantstorage/drives-enclosures/index.html IBM : http://www-03.ibm.com/systems/storage/disk/hdd.html 2.4.4 Midline Enterprise Server A suitable server for data backup should focus more on its storage capability, although we can not disregard the processing power and the memory at most. It should accommodate enough and suitable storage media which can be see from the availability of the storage controller (RAID controller) ,the type and number of the hard disk slots. Table 11 provides example of branded midline enterprise server with range of suitable budget. 45 HP ProLiant DL 380 G6 HP ProLiant DL380 G7 IBM x3550 M2 IBM x5550 M3 Processor family Intel® Xeon® 5500 series, Intel® Xeon® 5600 series Intel® Xeon® 5600 series Intel® Xeon® 5600 series Intel® Xeon® 5500 series Memory type PC3-10600R RDIMMs DDR3 or PC310600E UDIMMs DDR3 PC3-10600R RDIMMs DDR3 or PC310600E RDIMMs DDR3 RDIMMs DDR3 or UDIMMs DDR3 RDIMMs DDR3 or UDIMMs DDR3 Maximum drive bays (16) SFF SAS/SATA with optional second drive cage (16) SFF SAS/SATA with optional second drive cage (8) SFF SAS, SATA, SSD (8) SFF SAS, SATA Storage controller Smart Array P410i Integrated Smart Array P410i Integrated Hardware RAID-0, -1, -1E or RAID-0, -1, -10, -5, -50 (with additional option -6,-60) model dependent Hardware RAID-0, -1, -1E or RAID-0, -1, -10 or RAID-0, -1, -10, -5, -50 with 256 MB or 512 MB cache (additional option RAID-6, -60 Power supply 460 Watt hot plug 460 Watt hot plug 1/2; 675 W each 1/2; 675 W each Table 11: Example of midline enterprise server specifications HP : http://h10010.www1.hp.com/wwpc/us/en/sm/WF04a/15351-15351-3328412-241644-241475.html IBM : http://www-03.ibm.com/systems/x/hardware/rack/index.html 46