Techweb has new articles up;
Trek To Central Storage
By Chuck Moozakis
Centralized management, universal access. These two concepts reflect the goal many IT managers have for their data storage needs. But getting there is quite a challenge.
The obstacles are varied and complex, ranging from the deployment of multiple operating systems to 24-hour, seven-day operating requirements. At the same time, the amount-and types-of data continue to explode, adding more stress to the network.
The only way to control this traffic, IT managers say, is to craft a system in which administration is centrally controlled, even as information is delivered over myriad systems and platforms.
"Centralized management of storage is a given," says market watcher Farid Neema, president of Peripheral Concepts Inc. "We have talked a lot about this as being something new, but it has never really left. What we are seeing is centralized management of a semidistributed storage environment. Instead of gigabytes, now we are talking about terabytes, but this data still needs to be available on a departmental basis. Although the departments have to be connected, the management of that data has to be centralized."
Consider Dean Powell, for example. Powell, manager of divisional support services for Missouri's department of revenue, is converting all of the Show Me State's paper-based tax records to digital format.
"We figure we have 20 million pages a year to convert and that will continue for the foreseeable future," Powell explains. "The goal is to have this information accessible to about 500 employees." His chief challenge: designing a platform in which data can be retrieved as quickly as possible while centralized control over the data is retained.
Then there is Mike Zanga, senior engineer at Greenwich Capital Markets, a unit of $11.5 billion National Westminster Bank plc. Zanga is crafting a hybrid Unix/NT network-wedded in part by Fibre Channel-that will provide Greenwich's 700-plus users with the data they need to serve customers and keep abreast of Wall Street developments.
"We have 300 gigabytes of data now and we will grow another 500 gigabytes within the next six months," Zanga says. Keeping tabs on the ballooning number of storage resources necessary to store all this data while making sure that disk and tape space is used most efficiently is his primary concern.
Data Explosion
The issues facing Powell and Zanga are only two of a growing number of challenges confronting IT managers as they assess data storage. But while managers' storage issues may differ, there is little debate over what's causing the crunch.
Simply put, there's been an explosion in the amount and types of information companies generate and store. Data is not only flowing in from new deployments such as data warehousing, corporate intranets and the World Wide Web, it's changing in context as well.
"Data is coming from everywhere," says Pete Koliopoulos, vice president of sales and marketing at tape-library vendor Emass Inc. "Managers are asking for more data; they want access to trending data. Employees are also grabbing more and more information from the Internet and intranet and want to hold onto it for a longer time. Most of that data is now text-based, but that will change, and that type of data will pile up over time."
Data also is being generated everywhere, thanks to the proliferation of branch offices and remote sites. The result is a growing mountain of data, all of which must be stored, backed up and made accessible.
To gain control, data administrators are using a variety of tactics, ranging from more powerful servers to faster and more sophisticated storage networks. But as gigabytes turn into terabytes, data administrators are expanding their arsenals to stay ahead of the crunch.
Missouri's Powell designed his storage and retrieval system around two Compaq file servers and a Hewlett-Packard optical storage jukebox. A 160-gigabyte array attached to one of the servers acts as a cache for frequently or recently requested documents. The result is that information is sent to customer service reps within 20 seconds if retrieved from the jukebox, and within 3 seconds if accessed from the cache. "The retrieval time is attractive," Powell says. "A lot of thought went into designing this network. It's all expandable, and we will grow with it."
Greenwich's Zanga solved his storage-monitoring problems with Storage Resource Manager, a resource-mapping application released by HighGround Systems Inc. "The question is how much storage do I have, because storage doesn't necessarily correspond to how many servers are on the network," he says.
The application lets Zanga determine how much space is being used and where bottlenecks are likely. "People used to throw up a rack and that was good enough for a year or so. But with all the data that's being created, combined with having to know where that data is, that doesn't cut it anymore," he says.
If the past few months are any indication, vendors are tapping into that mind-set. Many storage suppliers have outlined products and strategies designed to let administrators oversee how data is to be routed and stored. Consider the following:
- Sun Microsystems introduced its Intelligent Storage Architecture and StorEdge line of products, designed to go beyond Sun's Unix hallmark. Sun's ISA lets multiplatform devices work together by mapping applications to appropriate storage devices. Data supporting online transaction processing (OLTP) would be stored in a cache-equipped array; a data-warehousing app would be routed to a Fibre array to take advantage of Fibre's throughput. A forthcoming Java-enabled management suite will let administrators control data and storage devices.
- IBM bolstered its Seascape storage systems line by adding Web caching, additional cross-platform support and a new product, IBM InfoSpeed. InfoSpeed will let managers transfer information from S/390 systems to Unix and NT servers. Improvements to its Adstar Distributed Storage Manager application extend support to HP-UX and Sun Solaris environments. ADSM also will sport a new Java Web console that permits centralized administration of storage networks.
- Two of the biggest vendors, EMC Corp. and StorageTek Corp., each outlined plans to boost NT support as part of a move to deliver multiplatform services. EMC said it would offer an NT module for its Open Systems software; StorageTek earlier this month took the wraps off an entry-level disk/tape array aimed at the NT marketplace.
- And finally, Amdahl Corp. outlined a multiplatform strategy of its own, as it moves from the mainframe world to open systems.
Fibre Speeds
The technology driving these announcements is Fibre Channel, which moves data to disparate devices at blazingly fast speeds-up to 100 megabytes per second.
"Fibre provides the scalability that will be required to support the tens and hundreds of terabytes of storage infrastructure," says Dave Tang, director of business development at Fibre switch vendor Gadzoox Networks Inc. "With these large data stores, the ability to share data will be more and more important. Having multiple servers accessing a common storage network will facilitate this."
A common management level will be just as important, Tang explains. In a distributed environment, where servers and storage devices are scattered throughout the enterprise, the cost of managing storage can account for more than half an enterprise's storage budget. But in a centralized scenario, storage management costs nosedive, even as the amount of data administered increases.
"There is a very strong movement toward recentralization of data because of cost and security," Tang says. "The biggest challenge is multiplatform support."
In fact, multiplatform support is a primary goal of Fibre vendors and the Fibre Channel Association. "Ease of deployment and some interoperability issues do remain," says Joe Colgate, vice president of sales and marketing at Vixel Corp. "What we want to see is an open storage connection to link heterogeneous servers."
The ability to connect servers will likely come via switched-fabric topologies, where data is routed among hundreds of workstations and storage repositories. Ancor Communications Inc., Brocade Communications Systems, McData Corp. and Vixel are among the suppliers marketing Fibre switches, which will bring to storage-area networks the same flexibility used by the public switched telephone network to route voice and data traffic.
Fibre is only one weapon administrators will likely use to manage data. Other trends, including Web caching, more intelligent RAID and more sophisticated network-attached storage devices, also are gathering steam.
"Data storage and retrieval are enabling technologies," says Cheena Srinivasan, director of product marketing at Network Appliance Inc., a Web caching and storage vendor. "Caching is a very important ingredient."
With caching, companies can store often-requested Internet information closer to users, relieving bandwidth-pinched data pipes and reducing strain on storage resources. NetApp has already carved a niche with its NetCache device-introduced last December-and has scored more than $1 million in sales.
Meanwhile, many vendors are focusing on more intelligent RAID products. "There has been a shift on how companies are consuming arrays," says Mark J. Vargo, marketing manager at Data General Corp.'s Clariion unit. "Instead of putting a giant box on the floor and attaching servers to it, people want to buy storage for specific apps."
To meet those demands, Clariion late last year unveiled its Navisphere management app, which automatically routes data to specific storage arrays based on the data's application, regardless of platform.
"People already are highly distributed," Vargo says. "Now managers want tools that will let them manage remotely. If you have an OLTP server in Chicago and a data warehouse in London, they can now be controlled remotely from someone in New York. With multiple platforms, each one of these is different. Managers want to control all their attached arrays with a centralized tool."
That's the same tack taken by Tricord Systems with its forthcoming TSMS RAID management app. The application will let companies store data across multiple disk arrays by moving file intelligence to the storage system controller and then further distributing this intelligence across multiple controllers. The result is a single, shared data pool that can be configured on the fly.
Even as vendors trot out new storage products and technologies, IT managers are likely to be demanding even more robust solutions before too long. Remember that saying about death and taxes being the only sure things? IT managers had better add the creation of more data to that list as well. The good news is there's no shortage of reliable solutions to help on the trek to central storage.Trek To Central Storage
By Chuck Moozakis
Centralized management, universal access. These two concepts reflect the goal many IT managers have for their data storage needs. But getting there is quite a challenge.
The obstacles are varied and complex, ranging from the deployment of multiple operating systems to 24-hour, seven-day operating requirements. At the same time, the amount-and types-of data continue to explode, adding more stress to the network.
The only way to control this traffic, IT managers say, is to craft a system in which administration is centrally controlled, even as information is delivered over myriad systems and platforms.
"Centralized management of storage is a given," says market watcher Farid Neema, president of Peripheral Concepts Inc. "We have talked a lot about this as being something new, but it has never really left. What we are seeing is centralized management of a semidistributed storage environment. Instead of gigabytes, now we are talking about terabytes, but this data still needs to be available on a departmental basis. Although the departments have to be connected, the management of that data has to be centralized."
Consider Dean Powell, for example. Powell, manager of divisional support services for Missouri's department of revenue, is converting all of the Show Me State's paper-based tax records to digital format.
"We figure we have 20 million pages a year to convert and that will continue for the foreseeable future," Powell explains. "The goal is to have this information accessible to about 500 employees." His chief challenge: designing a platform in which data can be retrieved as quickly as possible while centralized control over the data is retained.
Then there is Mike Zanga, senior engineer at Greenwich Capital Markets, a unit of $11.5 billion National Westminster Bank plc. Zanga is crafting a hybrid Unix/NT network-wedded in part by Fibre Channel-that will provide Greenwich's 700-plus users with the data they need to serve customers and keep abreast of Wall Street developments.
"We have 300 gigabytes of data now and we will grow another 500 gigabytes within the next six months," Zanga says. Keeping tabs on the ballooning number of storage resources necessary to store all this data while making sure that disk and tape space is used most efficiently is his primary concern.
Data Explosion
The issues facing Powell and Zanga are only two of a growing number of challenges confronting IT managers as they assess data storage. But while managers' storage issues may differ, there is little debate over what's causing the crunch.
Simply put, there's been an explosion in the amount and types of information companies generate and store. Data is not only flowing in from new deployments such as data warehousing, corporate intranets and the World Wide Web, it's changing in context as well.
"Data is coming from everywhere," says Pete Koliopoulos, vice president of sales and marketing at tape-library vendor Emass Inc. "Managers are asking for more data; they want access to trending data. Employees are also grabbing more and more information from the Internet and intranet and want to hold onto it for a longer time. Most of that data is now text-based, but that will change, and that type of data will pile up over time."
Data also is being generated everywhere, thanks to the proliferation of branch offices and remote sites. The result is a growing mountain of data, all of which must be stored, backed up and made accessible.
To gain control, data administrators are using a variety of tactics, ranging from more powerful servers to faster and more sophisticated storage networks. But as gigabytes turn into terabytes, data administrators are expanding their arsenals to stay ahead of the crunch.
Missouri's Powell designed his storage and retrieval system around two Compaq file servers and a Hewlett-Packard optical storage jukebox. A 160-gigabyte array attached to one of the servers acts as a cache for frequently or recently requested documents. The result is that information is sent to customer service reps within 20 seconds if retrieved from the jukebox, and within 3 seconds if accessed from the cache. "The retrieval time is attractive," Powell says. "A lot of thought went into designing this network. It's all expandable, and we will grow with it."
Greenwich's Zanga solved his storage-monitoring problems with Storage Resource Manager, a resource-mapping application released by HighGround Systems Inc. "The question is how much storage do I have, because storage doesn't necessarily correspond to how many servers are on the network," he says.
The application lets Zanga determine how much space is being used and where bottlenecks are likely. "People used to throw up a rack and that was good enough for a year or so. But with all the data that's being created, combined with having to know where that data is, that doesn't cut it anymore," he says.
If the past few months are any indication, vendors are tapping into that mind-set. Many storage suppliers have outlined products and strategies designed to let administrators oversee how data is to be routed and stored. Consider the following:
- Sun Microsystems introduced its Intelligent Storage Architecture and StorEdge line of products, designed to go beyond Sun's Unix hallmark. Sun's ISA lets multiplatform devices work together by mapping applications to appropriate storage devices. Data supporting online transaction processing (OLTP) would be stored in a cache-equipped array; a data-warehousing app would be routed to a Fibre array to take advantage of Fibre's throughput. A forthcoming Java-enabled management suite will let administrators control data and storage devices.
- IBM bolstered its Seascape storage systems line by adding Web caching, additional cross-platform support and a new product, IBM InfoSpeed. InfoSpeed will let managers transfer information from S/390 systems to Unix and NT servers. Improvements to its Adstar Distributed Storage Manager application extend support to HP-UX and Sun Solaris environments. ADSM also will sport a new Java Web console that permits centralized administration of storage networks.
- Two of the biggest vendors, EMC Corp. and StorageTek Corp., each outlined plans to boost NT support as part of a move to deliver multiplatform services. EMC said it would offer an NT module for its Open Systems software; StorageTek earlier this month took the wraps off an entry-level disk/tape array aimed at the NT marketplace.
- And finally, Amdahl Corp. outlined a multiplatform strategy of its own, as it moves from the mainframe world to open systems.
Fibre Speeds
The technology driving these announcements is Fibre Channel, which moves data to disparate devices at blazingly fast speeds-up to 100 megabytes per second.
"Fibre provides the scalability that will be required to support the tens and hundreds of terabytes of storage infrastructure," says Dave Tang, director of business development at Fibre switch vendor Gadzoox Networks Inc. "With these large data stores, the ability to share data will be more and more important. Having multiple servers accessing a common storage network will facilitate this."
A common management level will be just as important, Tang explains. In a distributed environment, where servers and storage devices are scattered throughout the enterprise, the cost of managing storage can account for more than half an enterprise's storage budget. But in a centralized scenario, storage management costs nosedive, even as the amount of data administered increases.
"There is a very strong movement toward recentralization of data because of cost and security," Tang says. "The biggest challenge is multiplatform support."
In fact, multiplatform support is a primary goal of Fibre vendors and the Fibre Channel Association. "Ease of deployment and some interoperability issues do remain," says Joe Colgate, vice president of sales and marketing at Vixel Corp. "What we want to see is an open storage connection to link heterogeneous servers."
The ability to connect servers will likely come via switched-fabric topologies, where data is routed among hundreds of workstations and storage repositories. Ancor Communications Inc., Brocade Communications Systems, McData Corp. and Vixel are among the suppliers marketing Fibre switches, which will bring to storage-area networks the same flexibility used by the public switched telephone network to route voice and data traffic.
Fibre is only one weapon administrators will likely use to manage data. Other trends, including Web caching, more intelligent RAID and more sophisticated network-attached storage devices, also are gathering steam.
"Data storage and retrieval are enabling technologies," says Cheena Srinivasan, director of product marketing at Network Appliance Inc., a Web caching and storage vendor. "Caching is a very important ingredient."
With caching, companies can store often-requested Internet information closer to users, relieving bandwidth-pinched data pipes and reducing strain on storage resources. NetApp has already carved a niche with its NetCache device-introduced last December-and has scored more than $1 million in sales.
Meanwhile, many vendors are focusing on more intelligent RAID products. "There has been a shift on how companies are consuming arrays," says Mark J. Vargo, marketing manager at Data General Corp.'s Clariion unit. "Instead of putting a giant box on the floor and attaching servers to it, people want to buy storage for specific apps."
To meet those demands, Clariion late last year unveiled its Navisphere management app, which automatically routes data to specific storage arrays based on the data's application, regardless of platform.
"People already are highly distributed," Vargo says. "Now managers want tools that will let them manage remotely. If you have an OLTP server in Chicago and a data warehouse in London, they can now be controlled remotely from someone in New York. With multiple platforms, each one of these is different. Managers want to control all their attached arrays with a centralized tool."
That's the same tack taken by Tricord Systems with its forthcoming TSMS RAID management app. The application will let companies store data across multiple disk arrays by moving file intelligence to the storage system controller and then further distributing this intelligence across multiple controllers. The result is a single, shared data pool that can be configured on the fly.
Even as vendors trot out new storage products and technologies, IT managers are likely to be demanding even more robust solutions before too long. Remember that saying about death and taxes being the only sure things? IT managers had better add the creation of more data to that list as well. The good news is there's no shortage of reliable solutions to help on the trek to central storage. |
