Riverbed WAN optimization delivered an affordable and easy-to-deploy solution that has ?exceeded expectations.?
Energy and Utilities
Long-distance concurrent engineering
15 Steelhead appliances installed around Québec
Steelhead Mobile for mobile workforce and smaller partners
CATIA file download time slashed by nearly 90%
LAN-like CATIA performance long distances from Montreal
Concurrent engineering adopted for about 50% less than other solutions
$1.5 million dollar cost avoidance
Office applications run 10 to 25 times faster over the WAN
90% average data reduction across the network
Concurrent Engineering is Possible over a WAN, with Riverbed
Hydro-Québec generates, transmits and distributes electricity produced primarily from green sources. Hydroelectricity is the primary source, accounting for 98 percent of the company’s output. Hydro-Québec operates 58 hydroelectric power plants, a nuclear power plant, four conventional thermal stations, and a wind farm, for an installed capacity of 36.3 gigawatts. The company has more than 23,000 employees.
Hydro-Québec uses the extended enterprise model which, in a province of the size of Québec, means that many of the people who collaborate on a project – Hydro employees as well as the company’s engineering, procurement and construction (EPC) partners – work at some distance from the main office in Montreal. Some colleagues are located more than 1,000 kilometers (661 miles) away.
Challenge: Concurrent engineering’s large files
The size of the province presented an obstacle to Hydro-Québec’s attempts to adopt the practice of concurrent engineering, where many different parties work simultaneously on a single digital model to create a digital mock-up of a facility. Because the design has been validated in software, the construction process typically has fewer errors such as interferences.
Hydro-Québec uses the CAD application, CATIA, from Dassault Systèmes to create digital models. It uses the SmarTeam product lifecycle management (PLM) system, also from Dassault Systèmes, to manage CATIA data and engineering workflows. The obstacle Hydro-Québec ran into when trying to implement concurrent engineering was extremely slow performance when moving large CATIA files over network connections to the remote sites. Hydro’s digital mock-up files are up to 600 megabytes in size. In a test with a stopwatch, even a relatively small CATIA file of only 14 megabytes took 15 minutes to download over a typical 10-megabits/second connection with 15ms latency.
Real-time collaboration using the digital mock-ups, one of the main benefits of concurrent engineering, was impossible. In addition, the people in the remote offices couldn’t even work with CATIA without 30- to 60-second delays between every mouse click. This is because they had to access a digital catalog of standard CATIA parts and icons (a database of approximately 2,000 small files) as they worked, and that information was stored on a server in Montreal. “Response time was ugly,” says Daniel Brisebois, Hydro-Québec IT advisor and a 3D technical team leader.
Not ready to give up on the promise of concurrent engineering, Brisebois and his colleagues looked into a number of options for making it work over their wide area network (WAN). One was to deploy and maintain dedicated servers running CATIA and SmarTeam licenses at each partner’s site. “This option was very costly and could have created problems related to data integrity and synchronization, as well as putting our intellectual property at risk,” says Brisebois.
“Replication between servers could have resolved some of these issues, but the increased maintenance and support costs made this especially undesirable. In addition, this method only supported SmarTeam environments, and we use other applications such as Oracle, Microsoft Exchange, and LiveLink with our partners and remotes sites as well.”
Other options included providing access to the remote sites via Hewlett-Packard’s Remote Graphics Software (RGS), using the Citrix ICA client, and WAN optimization. The architects of the RGS solution requested bandwidth on the order of 40 megabits/second per user for the responsiveness needed by the end users. Given the distance between certain sites, it was impossible to establish that much bandwidth, not to mention the latency that would have greatly hampered this option’sviability. As for the Citrix ICA solution, it did not perform adequately enough to support the installation of CATIA and SmarTeam.
Solution: Affordable, easy-to-deploy WAN optimization
WAN optimization offered the greatest promise, although only if Hydro-Québec could find a solution that was easy for its partners to deploy. “They have their own networks and requirements, so it was important to have a product would be easy for them to install,” says André Gagnon, a senior architect at Hydro-Québec who developed the guidelines for the WAN optimization solution.
In conjunction with Hydro-Québec’s engineering department, Gagnon identified a total of 35 requirements that the WAN optimization solution would have to meet to ensure efficient collaboration across the extended enterprise. Other requirements included: a low cost per remote site to allow a rapid ROI for the smaller partners, support for legacy engineering systems, data integrity, and support for more than 30 connections per site.
Some research into the available options led the company to Riverbed Technology, whose WAN optimization solution has been rated highly by industry analysts. “Riverbed was judged by one analyst to have the best ability to execute and the best vision,” says Brisebois.
Benefits: Fewer design and engineering errors; faster engineering cycle
After a proof-of-concept that evaluated Riverbed Steelhead appliances at two sites, Hydro-Québec deployed 15 Steelhead appliances in the offices of its major partners around the province. For its own mobile workforce and smaller partners, the company is also beginning to use Steelhead Mobile software, which provides WAN optimization for laptops. (This solution will be more largely deployed within Hydro as soon as the company upgrades to 64-bit Windows 7. According to Brisebois, the laptops really need a 64-bit environment when working with those huge CATIA digital mock-ups.)
As soon as the Steelhead appliances were installed, the slow performance of office applications over the WAN became a thing of the past. “We installed the appliances and that acceleration happened automatically,” explains Brisebois. Riverbed has the greatest number of application- specific optimizations of any vendor. “By addressing issues of bandwidth, latency, and chatty protocols, the Steelhead appliances make applications such as Oracle, Citrix and LiveLink run between 10 to 25 times faster over the network than they did previously,” he adds.
But would this solution deliver on the promise of doing concurrent engineering over a WAN? “It was a little more challenging to get the Steelhead appliances to work with SmarTeam (where the CATIA data is stored),” Brisebois adds. “We had Riverbed’s help with that, and it works even better than we had hoped.”
The Steelhead appliances give everyone efficient access to the CATIA and SmarTeam data no matter where they are located. Remote access to this data is now about the same as local access. A 17-megabyte CATIA file that took 1 minute 25 seconds to download previously downloaded in 8 seconds after the Riverbed solution was deployed. Another way of looking at the difference is to note the enormous reduction in data that now travels over the WAN as this information is shared.
In one example, 48.6 gigabytes worth of information in the SmarTeam vault was sent to a remote site, but due to Riverbed’s ability to streamline data, only 2.2 gigabytes was actually transferred. Also, a remote construction site achieved more than 500 megabits/second throughput over a 10-megabit link – a 50-fold performance increase. “This performance was achieved when we upgraded our PLM system to the new release by prepopulating the cache and the file system of the remote appliances,” says Brisebois. “Without the Riverbed technology, we wouldn’t have been able to deploy the solution at all in the small time frame we had.”
Hydro-Québec takes advantage of the Riverbed Proxy File System (PFS) to replicate its CATIA catalog in read-only mode. This way, catalog data is stored on the Steelhead appliance itself, making that information available almost instantly to remote users. Now, instead of waiting up to a minute between mouse clicks, engineers are able to work with CATIA as if all of the catalog data were stored at their own facilities. And because the Steelhead compresses this data by about 97 percent, the catalog is now synchronized nightly, instead of monthly as in the past.
According to Émil Dobrescu, head of the 3D team at Hydro-Québec, “The stakeholders can now collaborate easily and efficiently, and procedural errors have decreased.” One of Hydro-Québec’s first projects to benefit from concurrent engineering was a 1,500-megawatt hydroelectric facility located on the Romaine River. Benefits included fewer engineering problems due to document mismatches, fewer errors caused by working with outdated files, a faster engineering cycle, and enhanced data integrity.
Hydro-Québec has seen significant cost savings from the Riverbed solution. Brisebois estimates that choosing Riverbed WAN optimization allowed his company to implement concurrent engineering for about 50 percent less than it would have cost if they had gone with another approach (buying dedicated servers and more CATIA licenses, for example). He also estimates that through consolidation (maintaining only a single server in the home office), Hydro-Québec has saved as much as $1.5 million on avoided IT costs.
Hydro-Québec wanted to practice concurrent engineering but wasn’t able to make it work due to the long distances between its headquarters and partners’ offices throughout the province. The problem: transferring CATIA and SmarTeam data over a WAN was too slow to permit real- time collaboration. Even day-to-day use of CATIA at the remote sites was agonizingly slow.
Some possible work-arounds were too expensive (dedicated CATIA servers at each site) or didn’t work satisfactorily (remote desktop/thin client). Riverbed WAN optimization delivered an affordable and easy-to-deploy solution that has “exceeded expectations.” With 15 Steelhead appliances now installed around Québec, CATIA response time at remote sites is similar to that in the Montreal headquarters where the software is housed.
Very large CATIA digital mock-up files now transfer fast enough to permit concurrent engineering. The Riverbed solution cost about half what the company might have spent on other approaches, and consolidation has accounted for $1.5 million in avoided costs as well.
The Riverbed solution has exceeded our expectations.
“For us, concurrent engineering is not workable over a WAN without Riverbed.”