Hard Disk Features
| To make the best decision in purchasing a hard disk for your system or to understand what distinguishes one brand of hard disk from another, you must consider many features. This section examines some of the issues you should consider when you evaluate drives:
CapacityAs stated earlier, a corollary of Parkinson's famous "law" can be applied to hard drives: "Data expands so as to fill the space available for its storage." This of course means that no matter how big a drive you get, you will find a way to fill it. If you've exhausted the space on your current hard disk, you might be wondering, "How much storage space is enough?" Because you are more likely to run out of space than have too much, you should aim high and get the largest drive that will fit within your budget. Modern systems are used to store many space-hungry file types, including digital photos, music, video, newer operating systems, applications, and games. As an example, according to hard drive manufacturer Western Digital, storing 600 high-res photos (500KB each), 12 hours of digital music, 5 games, 20 applications, and just 90 minutes of digital video requires an estimated 43GB of space. Although most drives today do typically hold that much data with room to spare, it might exceed the capacity of hard drives installed in many low-end systems sold at retail, especially after you factor in space used by the operating system and other installed programs. Running out of space causes numerous problems in a modern system, mainly because Windows, as well as many newer applications, uses a large amount of drive space for temporary files and virtual memory. When Windows runs out of room, system instability, crashes, and data loss are inevitable. Capacity LimitationsHow big a hard drive you can use depends somewhat on the interface you choose. Although the ATA interface is by far the most popular interface for hard drives, SCSI interface drives are also available. Each has different limitations, but those of ATA have always been lower than those of SCSI. When ATA was first created in 1986, it had a maximum capacity limitation of 137GB (65,536x16x255 sectors). BIOS issues further limited capacity to 8.4GB in systems earlier than 1998, and 528MB in systems earlier than 1994. Even after the BIOS problems were resolved, however, the 137GB limit of ATA remained. Fortunately, this was broken in the ATA-6 specification drafted in 2001. ATA-6 augments the addressing scheme used by ATA to allow drive capacity to grow to 144PB (petabytes, or quadrillion bytes), which is 248 sectors. This has opened the door allowing ATA drives over 137GB to be released. Obviously any drives larger than 137GB would by nature conform to ATA-6. However, if you are installing a drive larger than that, you should also ensure that your motherboard BIOS has ATA-6 support. BIOS LimitationsIf your current hard drive is 8GB or smaller, your system might not be capable of handling a larger drive without a BIOS upgrade because many older (pre-1998) BIOSs can't handle drives above the 8.4GB limit, and others (pre-2002) have other limits such as 137GB. Most ATA hard drives ship with a setup disk containing a software BIOS substitute such as Ontrack's Disk Manager or Phoenix Technologies' EZ-Drive (Phoenix purchased EZ-Drive creator StorageSoft in January 2002), but I don't recommend using a software BIOS replacement. Ez-Drive, Disk Manager, and their OEM offshoots (Drive Guide, MAXBlast, Data Lifeguard, and others) can cause problems if you need to boot from floppy or CD media or if you need to repair the nonstandard master boot record these products use. If your motherboard ROM BIOS dates before 1998 and is limited to 8.4GB or dates before 2002 and is limited to 137GB, and you want to install a larger drive, I recommend you first contact your motherboard (or system) manufacturer to see whether an update is available. Virtually all motherboards incorporate a flash ROM, which allows for easy updates via a utility program. Internal ATA drives larger than 137GB require 48-bit logical block address (LBA) support. This support must be provided in the operating system; it can also be provided in the BIOS, or both. It is best if both the OS and the BIOS support it, but it can be made to work if only the OS has the support. 48-bit LBA support in the OS requires
These cards use an ISA slot and contain a flash ROM that adds 48-bit LBA support for your existing ATA host adapters. These cards are ROM only. They don't have onboard ATA interfaces, and your drives remain connected to your previously existing ATA host adapters (normally built in to the motherboard). If you have both OS and BIOS support for 48-bit LBA, you can simply install and use the drive like any other internal drive. On the other hand, if you do not have 48-bit LBA support in the BIOS, but you do have it in the OS, portions of the drive past 137GB are not recognized or accessible until the OS is loaded. This means that if you are installing the OS to a blank hard drive and booting from an original XP (pre-SP1) CD or earlier, you need to partition up to the first 137GB of the drive at installation time. After the OS is fully installed and the service packs added, the remainder of the drive beyond 137GB is recognized. At that point, you can then either partition the remainder of the drive beyond 137GB using the XP Disk Management tools or use a third-party partitioning program such as PartitionMagic or Partition Commander to resize the first partition to use the full drive. If you are booting from an XP SP1 or later CD (meaning a CD with Service Pack 1 already applied), you can recognize and access the entire drive during the OS installation and partition the entire drive as a single partition greater than 137GB if you like. If you need more or faster ATA interface connections, or if you don't have ISA slots on your motherboard, you can use PCI-based add-on cards (such as the Ultra133 TX2 and Ultra100 TX2) from companies like Maxtor and Promise Technologies. These cards support drives up to and beyond the 137GB limit imposed by the ATA-5 and older standards. These cards also have two ATA host adapter interfaces onboard that support two drives each (four drives per card). These cards do support ATA-133 and ATA-100 interface speeds and are backward-compatible with older, slower ATA drives. Maxtor and Western Digital sell drives in kits bundled with an ATA-133 PCI host adapter card that also provides BIOS support for drives beyond 137GB. USB, FireWire, and SCSI drives don't have these capacity issues because they don't rely on the ROM BIOS for support and use OS managed drivers instead. SCSI was designed from the beginning with fewer limitations than ATA, which is why SCSI is more commonly used in high-performance file servers, workstations, and other high-performance computer systems. Even though SCSI originated prior to ATA, the architects had the foresight to allow SCSI to address devices up to 2.2TB (terabytes, or trillion bytes) in capacity (232 sectors). In 2001, the SCSI command set was further upgraded to support drives up to 9.44ZB (zettabytes, or sextillion bytes) in capacity (264 sectors). Because SCSI was initially designed with fewer limitations and greater performance in mind, manufacturers have traditionally released their largest capacity drives in SCSI versions first. With the advent of Serial ATA, however, the gap is quickly closing. Because of the changes in 2001 to both ATA and SCSI, it will be many years before the capacity limitations of either interface become a problem. Operating System LimitationsMore recent operating systems such as Windows Me, as well as Windows 2000 and XP, fortunately don't have any problems with larger drives. However, older operating systems might have limitations when it comes to using large drives. DOS generally does not recognize drives larger than 8.4GB because those drives are accessed using LBA and DOS versions 6.x and lower use only CHS addressing. Windows 95 has a 32GB hard disk capacity limit, and there is no way around it other than upgrading to Windows 98 or newer. Additionally, the retail or upgrade versions of Windows 95 (also called Windows 95 OSR 1 or Windows 95a) are further limited to using only the FAT16 (16-bit file allocation table) file system, which carries a maximum partition size limitation of 2GB. Therefore, if you had a 30GB drive, you would be forced to divide it into 15 2GB partitions, with each appearing as a separate drive letter (drives C:Q: in this example). Windows 95B and 95C can use the FAT32 file system, which allows partition sizes up to 2TB. Note that because of internal limitations, no version of FDISK can create partitions larger than 512MB. Windows 98 supports large drives, but a bug in the FDISK program included with Windows 98 reduces the reported drive capacity by 64GB for drives over that capacity. The solution is an updated version of FDISK that can be downloaded from Microsoft. Another bug appears in the FORMAT command with Windows 98: If you run FORMAT from a command prompt on a partition over 64GB, the size isn't reported correctly, although the entire partition is formatted. PerformanceWhen you select a hard disk drive, one of the important features you should consider is the performance (speed) of the drive. Hard drives can have a wide range of performance capabilities. As is true of many things, one of the best indicators of a drive's relative performance is its price. An old saying from the automobile-racing industry is appropriate here: "Speed costs money. How fast do you want to go?" The speed of a disk drive is typically measured in two ways:
Transfer RateThe transfer rate is probably more important to overall system performance than any other statistic, but it is also one of the most misunderstood specifications. The problem stems from the fact that several transfer rates can be specified for a given drive; however, the most important of these is usually overlooked. Don't be fooled by interface transfer rate hype, especially around ATA-133 or SATA-150. A far more important gauge of a drive's performance is the average media transfer rate, which is significantly lower than the interface rate of 133MBps or 150MBps. The media transfer rate represents the average speed at which the drive can actually read or write data. By comparison, the interface transfer rate merely indicates how quickly data can move between the motherboard and the buffer on the drive. The rotational speed of the drive has the biggest effect on the drive's true transfer speed; in general, drives that spin at 10,000rpm transfer data faster than 7,200rpm drives, and 7,200rpm drives transfer data faster than those that spin at 5,400rpm. If you are looking for performance, be sure to check the true media transfer rates of any drives you are comparing. The confusion results from the fact that drive manufacturers can report up to seven different transfer rates for a given drive. Perhaps the least important (but one that people seem to focus on the most) is the raw interface transfer rate, which for most modern ATA drives is either 100MBps or 133MBps, or 150MBps for Serial ATA drives. Unfortunately, few people seem to realize that the drives actually read and write data much more slowly than that. The more important transfer rate specifications are the media transfer rates, which express how fast a drive can actually read or write data. Media transfer rates can be expressed as a raw maximum, a raw minimum, a formatted maximum, formatted minimum, or averages of either. Few report the averages, but they can be easily calculated. The media transfer rate is far more important than the interface transfer rate because the media transfer rate is the true rate at which data can be read from (or written to) the disk. In other words, it tells how fast data can be moved to and from the drive platters (media). It is the rate that any sustained transfer can hope to achieve. This rate is usually reported as a minimum and maximum figure, although many drive manufacturers report the maximum only. Media transfer rates have minimum and maximum figures because drives today use zoned recording with fewer sectors per track on the inner cylinders than the outer cylinders. Typically, a drive is divided into 16 or more zones, with the inner zone having about half the sectors per track (and therefore about half the transfer rate) of the outer zone. Because the drive spins at a constant rate, data can be read twice as fast from the outer cylinders than from the inner cylinders. Another issue is the raw transfer rate versus the formatted transfer rate. The raw rate refers to how fast bits can be read off the media. Because not all bits represent data (some are intersector, servo, ECC, or ID bits), and because some time is lost when the heads have to move from track to track (latency), the formatted transfer rate represents the true rate at which user data can be read from or written to the drive. Note that some manufacturers report only raw internal media transfer rates, but you usually can calculate that the formatted transfer rates are about three-fourths of the raw rates. This is because the user data on each track is only about three-fourths of the actual bits stored due to servo, ECC, ID, and other overhead that is stored. Likewise, some manufacturers report only maximum transfer rates (either raw, formatted, or both); in that case, you generally can assume the minimum transfer rate is one-half of the maximum and that the average transfer rate is three-fourths of the maximum. Let's look at a specific drive as an example. The Hitachi Deskstar 120GXP is considered a fast ATA drive. It spins at 7,200rpm and supports the ATA/100 interface transfer rate (Ultra DMA Mode 5, which is 100MBps from the drive controller to the motherboard host adapter). As with all drives I know of, the actual (media) transfer rate is much less. Table 9.7 shows the specifications for the 7,200rpm Ultra-ATA/100 Hitachi (IBM) Deskstar 120GXP drive.
As you can see, the true transfer rate for this drive is between 57.02MBps and 27.53MBps, or an average of about 42.27MBpsless than half of the ATA/100 interface transfer rate. Of course, if this were your drive, you wouldn't be disappointed because 42.27MBps is excellent performance. In fact, this is one of the fastest ATA drives on the market. Many other ATA drives would have equal or slower performance. A common question I get is about upgrading the ATA interface in a system. Many people are using older motherboards that support only ATA/33 (Ultra DMA Mode 2) or ATA/66 (Ultra DMA Mode 4) modes and not the faster ATA/100 (Ultra DMA Mode 5), ATA/133 (Ultra DMA Mode 6), or SATA specifications. After studying the true formatted media transfer rates of most drives, you can see why I generally do not recommend installing a separate ATA/100 or ATA/133 host adapter for those systems, unless you need the additional host adapters to attach more drives. From a pure performance perspective, those who perform such an upgrade will most likely see little, if any, increase in performance. This is because in almost all cases, the drives they are using are on average slower than even ATA/33and often significantly slower than the ATA/66, ATA/100, or ATA/133 interface speeds. Two primary factors contribute to transfer rate performance: rotational speed and the linear recording density or sector-per-track figures. When comparing two drives with the same number of sectors per track, the drive that spins more quickly transfers data more quickly. Likewise, when comparing two drives with identical rotational speeds, the drive with the higher recording density (more sectors per track) is faster. A higher-density drive can be faster than one that spins fasterboth factors have to be taken into account. Let's look at another example. Similar to the IBM 120GXP, the Maxtor DiamondMax D540X-4G120J6 is also a 120GB ATA drive. It spins at 5,400rpm and supports the ATA/133 interface transfer rate (Ultra DMA Mode 6, which is 133MBps from the drive controller to the motherboard host adapter). Table 9.8 shows the specifications for the 5,400rpm Ultra-ATA/133 Maxtor DiamondMax D540X-4G120J6 120GB ATA drive.
As you can see, the true transfer rate for this drive is between 41.29MBps and 20.64MBps, or an average of about 30.97MBpsless than one-fourth of the interface transfer rate. Note the comparison between the two 120GB drives:
It is interesting to note that the drive with the faster interface transfer rate (133MBps versus 100MBps) is actually slower overall by a fairly large margin (about 37%). Because the drives have about the same average number of sectors per track, the differential in the transfer speed performance is mainly due to the 33% greater rotational speed in one drive over the other. If you were choosing among 120GB drives and looking for the highest performance, you should almost always choose the drive with the fastest media transfer rate. Even though it operates at a slower 100MBps (ATA/100) interface transfer speed, it actually reads and writes data 37% faster overall than the other drive, even though the other drive supports 133MBps interface transfers (ATA/133). As you can see from this example, the interface transfer speed is almost meaningless. In fact, because neither drive can actually transfer data faster than 66MBps (even from the outer cylinders), using an interface transfer rate higher than that does not help performance much. So, if you were thinking about getting a new motherboard or maybe a separate host adapter card for the sole purpose of increasing drive performance, save your money. To be fair, there will be a slight benefit to higher interface transfer speeds in that data from the buffer on the drive controller can be transferred to the motherboard at interface speed, rather than media speed. These buffers are usually 2MB or less and help only with repetitive transfers of small amounts of data. However, if you perform repetitive transfers frequently, drives with larger 8MB buffers can improve performance with applications that perform repetitive transfers. Refer to Table 9.7 for information on current drives with 8MB buffers. All other things being equal, a drive that spins faster transfers data faster, regardless of the interface transfer rate. Unfortunately, it is rare that all other things are exactly equal, so you should consult the drive specifications listed in the data sheet or manual for the drive to be sure. One of the fastest rotating drives today is the Seagate Cheetah X15, which spins at 15,000rpm. Table 9.9 shows the specifications for the 15,000rpm Ultra4-SCSI/320 Seagate Cheetah X15-36LP (ST-336732LW) drive.
As a comparison, Table 9.10 contains the specifications for the 10,000rpm Ultra4-SCSI/320 Seagate Cheetah 36ES (ST-336746LW) drive.
As you can see, although the 15,000rpm drive spins 50% faster, it actually transfers data only about 7% faster. It also costs about 38% more for the same capacity. Note that neither of these drives comes close to the Ultra4 SCSI (320MBps) bandwidth the interface would allow. One difference between ATA and SCSI, though, is that all the SCSI drives on a given channel can more effectively share the bandwidth. With a comparison such as this, you can see that you need to be careful. Don't just compare one specification, such as interface speed or rotational speed, because these can be misleading. The interface speed is relatively meaningless, and although the rotational speed is much more important, some drives have a slower media transfer rate than others even though they spin faster. Be careful with simplistic comparisons. With hard drives, the bottom line is that the media transfer rate is probably the most important specification you can know about a driveand faster is better. To find the transfer specifications for a given drive, look in the data sheet or preferably the documentation or manual for the drive. These can usually be downloaded from the drive manufacturer's website. This documentation often reports the maximum and minimum sector per track specifications, whichcombined with the rotational speedcan be used to calculate true formatted media performance. You should be looking for the true number of physical sectors per track for the outer and inner zones. Therefore, you should be aware that many drives (especially zoned-bit recording drives) are configured with sector translation, which means the number of sectors per track reported by the BIOS has little to do with the actual physical characteristics of the drive. You must know the drive's true physical parameters, rather than the values the BIOS uses. When you know the sector per track (SPT) and rotational speed figures, you can use the following formula to determine the true media data transfer rate in millions of bytes per second (MBps):
For example, the Hitachi Deskstar 120GXP drive spins at 7,200rpm and has an average of 688 sectors per track. The average media transfer rate for this drive is figured as follows:
Using this formula, you can calculate the media transfer rate of any drive if you know the rotational speed and average sectors per track. Average Seek TimeAverage seek time, usually measured in milliseconds (ms), is the average amount of time it takes to move the heads from one cylinder to another a random distance away. One way to measure this specification is to run many random track-seek operations and then divide the timed results by the number of seeks performed. This method provides an average time for a single seek. The standard method used by many drive manufacturers when reporting average seek times is to measure the time it takes the heads to move across one-third of the total cylinders. Average seek time depends only on the drive itself; the type of interface or controller has little effect on this specification. The average seek rating is primarily a gauge of the capabilities of the head actuator mechanism. Note Be wary of benchmarks that claim to measure drive seek performance. Most ATA/IDE and SCSI drives use a scheme called sector translation, so any commands the drive receives to move the heads to a specific cylinder might not actually result in the intended physical movement. This situation renders some benchmarks meaningless for those types of drives. SCSI drives also require an additional step because the commands first must be sent to the drive over the SCSI bus. These drives might seem to have the fastest access times because the command overhead is not factored in by most benchmarks. However, when this overhead is factored in by benchmark programs, these drives receive poor performance figures. LatencyLatency is the average time (in milliseconds) it takes for a sector to be available after the heads have reached a track. On average, this figure is half the time it takes for the disk to rotate once. A drive that spins twice as fast would have half the latency. Latency is a factor in disk read and write performance. Decreasing the latency increases the speed of access to data or files and is accomplished only by spinning the drive platters more quickly. Latency figures for most popular drive rotational speeds are shown in Table 9.11.
Many drives today spin at 7,200rpm, resulting in a latency time of only 4.17ms, whereas others spin at 10,000rpm or even 15,000rpm, resulting in incredible 3.00ms or 2.00ms latency figures. In addition to increasing performance where real-world access to data is concerned, spinning the platters more quickly also increases the data-transfer rate after the heads arrive at the desired sectors. Average Access TimeA measurement of a drive's average access time is the sum of its average seek time plus latency. The average access time is usually expressed in milliseconds. A measurement of a drive's average access time (average seek time plus latency) provides the average total amount of time required for the drive to access a randomly requested sector. Cache Programs and Caching ControllersAt the software level, disk cache programs such as SMARTDRV (DOS) and VCACHE (Windows) can have a major effect on disk drive performance. These cache programs hook into the BIOS hard drive interrupt and intercept the read and write calls to the disk BIOS from application programs and device drivers. When an application program wants to read data from a hard drive, the cache program intercepts the read request, passes the read request to the hard drive controller in the usual way, saves the data read from the disk in its cache memory buffer, and then passes the data back to the application program. Depending on the size of the cache buffer, data from numerous sectors can be read into and saved in the buffer. When the application wants to read more data, the cache program again intercepts the request and examines its buffers to see whether the requested data is still in the cache. If so, the program passes the data back from the cache to the application immediately, without another hard drive operation. Because the cached data is stored in memory, this method speeds access tremendously and can greatly affect disk drive performance measurements. Most controllers now have some form of built-in hardware buffer or cache that doesn't intercept or use any BIOS interrupts. Instead, the drive caches data at the hardware level, which is invisible to normal performance-measurement software. Manufacturers originally included track read-ahead buffers in controllers to permit 1:1 interleave performance. Some manufacturers now increase the size of these read-ahead buffers in the controller, whereas others add intelligence by using a cache instead of a simple buffer. Many ATA and SCSI drives have cache memory built directly into the drive's onboard controller. Most newer ATA drives have 2MB of built-in cache; many high-performance ATA drives have 8MB of cache. Most SCSI drives have 8MB while some have up to 16MB. I remember the days when 1MB or 2MB of RAM was a lot of memory for an entire system. Nowadays, some 3 1/2" hard disk drives can have up to 16MB of cache memory built right in! Although software and hardware caches can make a drive faster for routine or repetitive data transfer operations, a cache will not affect the true maximum transfer rate the drive can sustain. Interleave SelectionIn a discussion of disk performance, the issue of interleave often comes up. Although traditionally this was more a controller performance issue than a drive issue, modern ATA/IDE and SCSI hard disk drives with built-in controllers are fully capable of processing the data as fast as the drive can send it. In other words, all modern ATA and SCSI drives are formatted with no interleave (sometimes expressed as a 1:1 interleave ratio). On older hard drive types, such as MFM and ESDI, you could modify the interleave during a low-level format to optimize the drive's performance. Today, drives are low-level formatted at the factory and interleave adjustments are a moot topic. Note For more information on interleaving and cylinder skewing as used on older drives, see the sections "Interleave Selection" and "Head and Cylinder Skewing" in Chapter 10 of Upgrading and Repairing PCs, 12th Edition, included in its entirety on the disc accompanying this book. ReliabilityWhen you shop for a drive, you might notice a statistic called the mean time between failures (MTBF) described in the drive specifications. MTBF figures usually range from 300,000 to 1,000,000 hours or more. I usually ignore these figures because they are derived theoretically. In understanding the MTBF claims, you must understand how the manufacturers arrive at them and what they mean. Most manufacturers have a long history of building drives, and their drives have seen millions of hours of cumulative use. They can look at the failure rate for previous drive models with the same components and calculate a failure rate for a new drive based on the components used to build the drive assembly. For the electronic circuit board, they also can use industry-standard techniques for predicting the failure of the integrated electronics. This enables them to calculate the predicted failure rate for the entire drive unit. To understand what these numbers mean, you must know that the MTBF claims apply to a population of drives, not an individual drive. This means that if a drive claims to have an MTBF of 500,000 hours, you can expect a failure in that population of drives in 500,000 hours of total running time. If 1,000,000 drives of this model are in service and all 1,000,000 are running simultaneously, you can expect one failure out of this entire population every half-hour. MTBF statistics are not useful for predicting the failure of any individual drive or a small sample of drives. You also need to understand the meaning of the word failure. In this sense, a failure is a fault that requires the drive to be returned to the manufacturer for repair, not an occasional failure to read or write a file correctly. Finally, as some drive manufacturers point out, this measure of MTBF should really be called mean time to first failure. "Between failures" implies that the drive fails, is returned for repair, and then at some point fails again. The interval between repair and the second failure here would be the MTBF. Because in most cases, a failed hard drive that would need manufacturer repair is replaced rather than repaired, the whole MTBF concept is misnamed. The bottom line is that I do not really place much emphasis on MTBF figures. For an individual drive, they are not accurate predictors of reliability. However, if you are an information systems manager considering the purchase of thousands of PCs or drives per year or a system vendor building and supporting thousands of systems, it might be worth your while to examine these numbers and study the methods used to calculate them by each vendor. Most hard drive manufacturers designate their premium drives as Enterprise class drives, meaning they are designed for use in environments requiring full-time usage and high reliability and carry the highest MTBF ratings. If you can understand the vendor's calculations and compare the actual reliability of a large sample of drives, you can purchase more reliable drives and save time and money in service and support. SMARTSelf-Monitoring, Analysis, and Reporting Technology (SMART) is an industry standard providing failure prediction for disk drives. When SMART is enabled for a given drive, the drive monitors predetermined attributes that are susceptible to or indicative of drive degradation. Based on changes in the monitored attributes, a failure prediction can be made. If a failure is deemed likely to occur, SMART makes a status report available so the system BIOS or driver software can notify the user of the impending problems, perhaps enabling the user to back up the data on the drive before any real problems occur. Predictable failures are the types of failures SMART attempts to detect. These failures result from the gradual degradation of the drive's performance. According to Seagate, 60% of drive failures are mechanical, which is exactly the type of failures SMART is designed to predict. Of course, not all failures are predictable, and SMART can't help with unpredictable failures that occur without any advance warning. These can be caused by static electricity; improper handling or sudden shock; or circuit failure, such as thermal-related solder problems or component failure. SMART was originally created by IBM in 1992. That year IBM began shipping 3 1/2" hard disk drives equipped with Predictive Failure Analysis (PFA), an IBM-developed technology that periodically measures selected drive attributes and sends a warning message when a predefined threshold is exceeded. IBM turned this technology over to the ANSI organization, and it subsequently became the ANSI-standard SMART protocol for SCSI drives, as defined in the ANSI-SCSI Informational Exception Control (IEC) document X3T10/94-190. Interest in extending this technology to ATA drives led to the creation of the SMART Working Group in 1995. Besides IBM, other companies represented in the original group were Seagate Technology, Conner Peripherals (now a part of Seagate), Fujitsu, Hewlett-Packard, Maxtor, Quantum, and Western Digital. The SMART specification produced by this group and placed in the public domain covers both ATA and SCSI hard disk drives and can be found in most of the more recently produced drives on the market. The SMART design of attributes and thresholds is similar in ATA and SCSI environments, but the reporting of information differs. In an ATA environment, driver software on the system interprets the alarm signal from the drive generated by the SMART "report status" command. The driver polls the drive on a regular basis to check the status of this command and, if it signals imminent failure, sends an alarm to the operating system where it is passed on via an error message to the end user. This structure also enables future enhancements, which might allow reporting of information other than drive failure conditions. The system can read and evaluate the attributes and alarms reported in addition to the basic "report status" command. SCSI drives with SMART communicate a reliability condition only as either good or failing. In a SCSI environment, the failure decision occurs at the disk drive and the host notifies the user for action. The SCSI specification provides for a sense bit to be flagged if the drive determines that a reliability issue exists. The system then alerts the end user via a message. Note that traditional disk diagnostics such as Scandisk work only on the data sectors of the disk surface and do not monitor all the drive functions that are monitored by SMART. Most modern disk drives keep spare sectors available to use as substitutes for sectors that have errors. When one of these spares is reallocated, the drive reports the activity to the SMART counter but still looks completely defect-free to a surface analysis utility, such as Scandisk. Drives with SMART monitor a variety of attributes that vary from one manufacturer to another. Attributes are selected by the device manufacturer based on their capability to contribute to the prediction of degrading or fault conditions for that particular drive. Most drive manufacturers consider the specific set of attributes being used and the identity of those attributes as vendor specific and proprietary. Some drives monitor the floating height of the head above the magnetic media. If this height changes from a nominal figure, the drive could fail. Other drives can monitor different attributes, such as ECC circuitry that indicates whether soft errors are occurring when reading or writing data. Some of the attributes monitored on various drives include the following:
Each attribute has a threshold limit that is used to determine the existence of a degrading or fault condition. These thresholds are set by the drive manufacturer, can vary among manufacturers and models, and can't be changed. The basic requirements for SMART to function in a system are simple: You just need a SMART-capable hard disk drive and a SMART-aware BIOS or hard disk driver for your particular operating system. If your BIOS does not support SMART, utility programs are available that can support SMART on a given system. These include Norton Utilities from Symantec, EZ Drive from StorageSoft, and Data Advisor from Ontrack. When sufficient changes occur in the monitored attributes to trigger a SMART alert, the drive sends an alert message via an IDE/ATA or a SCSI command (depending on the type of hard disk drive you have) to the hard disk driver in the system BIOS, which usually reports the problem during the POST the next time the system boots. If you want more immediate reporting, you can run a utility that queries the SMART status of the drive, such as SMART Explorer by Adenix (www.adenix.net) or HDD Health by Panterasoft (www.panterasoft.com). The first thing to do if you receive a SMART warning is to back up all the data on the drive. I recommend you back up to new media and do not overwrite any previous backups you might have, just in case the drive fails before the new backup is complete. After backing up your data, what should you do? SMART warnings can be caused by an external source and might not actually indicate that the drive itself is going to fail. For example, environmental changes such as high or low ambient temperatures can trigger a SMART alert, as can excessive vibration in the drive caused by an external source. Additionally, electrical interference from motors or other devices on the same circuit as your PC can induce these alerts. If the alert was not caused by an external source, a drive replacement might be indicated. If the drive is under warranty, contact the vendor and ask whether they will replace it. If no further alerts occur, the problem might have been an anomaly, and you might not need to replace the drive. If you receive further alerts, replacing the drive is recommended. If you can connect both the new and existing (failing) drive to the same system, you might be able to copy the entire contents of the existing drive to the new one, saving you from having to install or reload all the applications and data from your backup. Because standard copy commands or drag-and-drop methods don't copy system files, hidden files, or files that are open, to copy an entire drive successfully and have the destination copy remain bootable, you need a special application such as Symantec Norton Ghost or Acronis Drive Image. CostThe cost of hard disk storage is continually falling. You can now purchase a 300GB ATA drive for around $100, which is about 1/30 of a cent per megabyte. A drive I bought in 1983 had a maximum capacity of 10MB and cost $1,800. At current pricing (0.03 cents per megabyte), that drive is worth about one-third of a penny! Of course, the cost of drives continues to fall, and we can expect even greater capacities and lower prices in the future. |
EAN: 2147483647
Pages: 283