If you cannot find the answers you are looking for in our site, and your question isn’t answered here, please let us know by email. We would like for this site to be as useful as possible. Thanks in advance for your input!
Streaming is the controlled delivery of a very large file from a server to a client. It’s just like downloading a piece of software or getting a file from FTP, except that the file being downloaded contains audio or video, and the streaming server and the streaming client use special software to negotiate the quality and rate of the download.
If the underlying network has problems during the download, the client and server software adjust the video on the fly to maintain as high a quality experience as possible. If they can’t negotiate, the result is seen as “buffering” or stuttering and halts in the video. This is far more perceptible in a video that you are watching in real time while you download it, than it is in a file that can download in the background while you continue to surf. This is why it is so important to use a real streaming server instead of just letting people download video from your web server.
Update as of Summer 2009: With the release of Silverlight 3, Flash 10, and the HTML5 video player built into Safari on the iPhone, Mac, and Windows, it's now possible to target most desktops and devices using a single H.264 encoded MPEG4 (.mp4) file. Our network both encodes and delivers this type of video, using each kind of video server and for each kind of video player. If Firefox decided to switch their HTML5 video tag support from .OGG to .MP4 as well, we'd all finally be able to forget about the "which formats?" question for a while!
Advection.NET recommends Windows Media or Silverlight for both audio and video, mainly for best cost at a given level of quality. It is still less expensive to encode, host, and maintain a Windows Media or Silverlight library than it is to deliver content in the other formats. Less cost means lower investment, and that lower investment means a higher return for the client. However, our streaming video delivery network works equally well with Windows Media, QuickTime, Flash, or Silverlight. Our Locus 3s appliance accelerates all types of large file delivery, which includes the main streaming formats, and most other types of large files such as HTTP downloads as well.
Yes, we specialize in Windows Media and Silverlight. Before Advection.NET, we built the world’s highest quality Windows Media network, and we brought that expertise with us. Part of that experience is in making sure our clients made money from their streaming. When using a platform other than Windows Media or Silverlight, we find more money has to be made just to break even.
We have many years experience managing and executing each portion of the media production and delivery process. For a client such as World Wrestling Entertainment (WWE), our founders provided camera crews, traveling digital video studios to do live streaming production on location in hundreds of venues around the USA, encoding, and of course, hosting of the live and packaged content.
We are among the leaders in delivering on-demand streaming, but we are experienced in going well beyond that, with Digital Rights Management or Pay-Per-View features to protect your content and help you derive immediate revenue from it. These are areas we’ve been focused on for a relatively long time in Internet terms. (See the Microsoft press releases noted in our Working with DRM page.)
We’ve been in the streaming business since 1998, and in the Internet business since 1993. Very few have as much experience at this as we do, and even fewer have consistently made money for their clients. Our show producers have degrees in broadcast journalism, and experience ranging from nightly news to cable pay-per-view; they’re not just college kids with a video camera. Our encoding specialists have encoded content for clients ranging from Fortune 50 executives to Hollywood studios, hand tailoring each stream to the content it contains. Finally, our engineers have built a next generation streaming server network from the ground up solely for streaming, to serve the highest quality streams to the broadest audience at the lowest delivery costs. Also, see the question below comparing a streaming specialist such as Advection.NET with a global backbone and services company such as Qwest.
If everything is running at maximum capacity, delivering streams to modem users, the capacity produced by our configurations in a lab is around 2,500,000 simultaneous streams, or approximately 27 billion streams a month. In the real world, this is probably closer to 250,000 simultaneous streams, and 2 billion streams a month. By way of comparison, the most popular streaming sites in the world deliver at most 20,000 simultaneous streams, or tens of millions of streams a month.
The streaming industry's dirty little secret is almost nobody in the industry measures the quality of their streaming. Really, the only way to measure success is by satisfied viewers, but that’s the last thing most streaming companies want to go by. The next best way is by stream start times or “failed” streams, as measured by third parties like Keynote. These are also shown by the streaming server logs and our proprietary streaming server plug-ins. We look for broken connections or excessive buffering, then attempt to isolate and resolve the reasons for this to minimize future recurrences. We also provide our customers with the raw logs, so they can see the performance for themselves.
Many streaming providers don’t worry about the difference between a network tuned for web hosting and a network tuned for streaming. Many providers have never delivered extreme volumes of streams, and therefore have little experience in tuning a network for high-volume low-cost streaming.
Our experience in delivering hundreds of millions of streams to the most demanding audiences, with real time monitoring and real time viewer feedback, has allowed us to design this network from the ground up to deliver a better stream. A key component of our network is our own streaming server optimizing appliance, a device that manages streaming traffic in real time to ensure the highest quality delivery to each viewer.
Put simply, each stream is delivered at the highest quality available. Fewer viewers will experience buffering or lost streams. More satisfied viewers equals more recurring visitors, and this drives revenue up throughout the web site, not just from streaming.
There is a lot of disagreement about whether live, mass-audience, entertainment events can pay off on the Internet today. We think if audiences will be reaching those levels without personalized video, the event is better off being on TV. Our goal is to be in business for the long haul. This means we want to work with customers who understand being in business requires that we make a profit, and who respect that our making a profit means we’ll still be around to service them long after the competition who is subsidizing business is gone. Of course, for customers to be happy with us making a profit, they need to make a profit too.
By distributing our streaming network across multiple Tier 1 backbone carriers, we are able to offer SLA’s that meet or exceed any single carrier’s. Please contact sales if you would like to receive a copy of the collocation or managed streaming agreements.
With five years experience delivering streaming media, we have learned the optimal configuration for the major streaming server platforms used today. However, we have a discussion on our video delivery network page of why today’s hardware needs help to deliver significant volumes of on-demand content. To get around these issues, our servers are managed by our Locus 3s appliance, which actively manages streaming traffic across the entire network of streaming servers, allowing each server to deliver several times more volume than it would on its own or even as part of a farm behind load balancers.
In Internet topology, geography is largely irrelevant. The key is to be as close as possible to as many users as possible from the standpoint of the electrons or photons flowing across the wires. However, in some cases, geography and Internet topology overlap, as in the case of major seaports being both physical and Internet gateways to regions on the other side of the ocean. This is why our primary two data centers happen to be in New York City and San Francisco (San Jose, actually). Because most Internet backbones make the trans-oceanic jump from these points, data centers in these locations have extremely high backbone peering and clean paths to both national and international users. Advection.NET has servers in four key peering points in North America, one so far in Europe, as well as a backup NOC for network management and data storage. Our key hubs are located in New York, NY; San Jose, CA; Washington, DC; Los Angeles, CA; Dallas, TX; Paris, France; and Frankfurt, Germany; our primary web hosting is in Phoenix, AZ; and our private “offsite” backup facility is near our corporate offices in Connecticut. (Check out the map of our network showing video delivery hubs and backbones.)
Yes. We have a complete customer service escalation process and a sizable engineering staff. It works like this:
Many reasons, with the more pertinent among them being responsiveness and expertise. Qwest, while a fine company, is also enormous; operating everything from one of the RBOC’s (US West), to data centers, to fiber, to networking, to managed services. This necessarily requires them to be great generalists, whatever their intent. Outsourcing, as we essentially do, two complex systems ourselves in the network and data center operation to other highly competent providers, allows us to focus our skills and attention more keenly on providing specific services — managed hosting with a specific emphasis on high-quality and reliable streaming.
Additionally, while we are a rapidly growing company, we have nothing like the number of customers of a Qwest. This is an enormous advantage to our customers in that while our expertise in our areas of operation is greater, the amount of focus we can provide to each customer is also much higher. Our team has previously realized these advantages for companies like WWF, Sony and Showtime.
Yes, at least as competitive. Our company buys bandwidth on an enormous scale — multiple gigabits ("Gbps") per datacenter. This ensures that bandwidth pricing is one of our major competitive advantages. Since we are not building the underlying backbones, but reselling the largest backbones in the world, we have very few “sunk costs”. If bandwidth prices in general continue to drop, we will continue to see savings even before others would be likely to. As a result, we can pass such savings along to our own customers. Scaling is likewise an easy issue for the same reason. We often price networking less expensively than the networks themselves even at volumes of 300 and 400 Mbps. Having this kind of purchasing power puts us in a unique position of being able not only to allow, but to facilitate our customers’ growth. Again, as one of the network builders, companies the size of Qwest have huge capital expenses on their books that we as a reseller do not. One has only to look at the headlines of recent years to see the problems associated with those expenses.
In this economy? But seriously, we're not here to burn money (we didn't take any VC money, for example), we're here to grow a profitable business. This is a function again of our unique position as a value-added provider which is able to rapidly implement technologies and grow without large capital expenses. It is further a function of the fact that each of our solutions has been custom tailored depending on the specific requirement. For this reason, our customers get what they need, and are able to make their own customers happy and grow as a result. Ensuring our customers’ growth will allow us to keep growing ourselves in a responsible, professional manner.
Yes. Provided we charge a fair, sustainable price for quality services, we can maintain moderately rapid growth, without seeking outside funding, indefinitely. Also, see the previous answer.
Because of the differences in the Russian educational system, Russian developers are extremely disciplined and extremely talented. You may have noticed that any time a major software company releases a new “content protection” or “digital rights management” scheme like Adobe’s e-Book software, it’s not long before Russian programmers have defeated the copy protection. In the areas of network security and content protection, these guys are exceptionally good. Rather than develop something here that’s hacked three days after release by Russians, we decided to work with Russian developers in the first place to make sure our software is robust and reliable. Also, Auragan’s founders are both of Russian heritage.
Pricing is usually provided as dollars per gigabyte transferred per month, and dollars per gigabyte stored per month. Many carriers have incremental costs depending on the number of locations you want them to use to cache your files. We can provide pricing for data, but to give you a more real world comparison, we'll describe the costs in terms of encoded media.
At the low end, our “Personal” plan at $19.95 per month will get you about 3,000 hours of FM quality delivery, and about 300 hours of FM quality storage. The same price gets you about 200 hours of TV quality delivery and about 20 hours of TV quality storage.
Our basic “Corporate” plan is $299.95 per month. This gets you nearly 4,000 hours of TV video delivery or 60,000 hours of FM audio delivery, and 400 hours of TV video storage or 6,000 hours of FM audio storage.
At the higher end, our “Media Plus” plan runs $9,995.00 per month. For this you receive a whopping 120,000 hours of DVD quality video delivery (double TV quality) or 1,000,000 hours of CD quality audio delivery (double FM quality). You can store 10,000 hours of DVD quality video or 100,000 hours of CD quality audio.
For accounts larger than this, we recommend moving to a "per terabyte transferred" tiered pricing model, with bandwidth and storage billed at incredibly competitive rates. When preparing this type of pricing, we will ask you about your ratio of gigabytes transferred per megabit sustained measured at 95th percentile. Check with your sales person for the latest figures.
For very large clients, it's not the total data transferred that matters, it's how large of a bandwidth commitment we need to make to our backbone providers. We get billed for bandwidth using the 95th percentile method. Instead of billing us for the amount of data we move, we get billed for the size of the pipe that moves that data. Have a look at this diagram:
The blue and pink shaded areas of the chart represent the amount of data, or gigabytes transferred, over the month, with weeks measured across the horizontal scale of the graph. The vertical scale on the graph represents the amount of bandwidth required to transfer that data during peak periods. Again, think of this whole chart as a pipe, think of the shaded areas as being filled with water, and think of the vertical spikes on this graph as the amount of water trying to go through the pipe on each day or week of the month. If someone is billing you for the water you send through the pipe, but they themselves pay for the size of the pipe they use, they will want to fit your water through the smallest pipe possible so they can make more money. They will probably talk about things like “capping” your bandwidth or other techniques of making sure you don’t get too big of a pipe. They’ll also try to fit as many other people’s water as possible through the same pipe, to use it all up. This is a recipe for network congestion.
Just like any major bandwidth provider, we pay for the size of the pipe. To make sure we can afford to give every customer the bandwidth they need to deliver streams smoothly, we base the billing for our largest customers on the size of pipe they need for their streams. Now, in the diagram above, you can see that at one point, the bandwidth (pipe required) hit 884 Kbps (or 0.884 megabits per second). That was the “peak” megabits sustained. But throughout the rest of the month, nothing near that was needed. Rather than bill the customer for the peak, it’s been figured out that by throwing away the top 5% of traffic, and billing for the remaining 95% of traffic, we can get a really good estimation of just how big a pipe you need to make sure you can always deliver streams smoothly. In this diagram, that’s at 377 Kbps (0.377 megabits per second). Sometimes you’ll need a little more, sometimes you’ll need a little less, but you can easily see how the area between that red “95th %ile” line and the bottom of the chart is the right size pipe to let all your blue and pink shaded data through.
Using our network monitoring software, we figure out your average bandwidth in every five minute period. We record each of these five minute averages on our usage database and plot them on your usage graph. By the end of the month, we have noted every five minute average. (If you’re curious, that’s about 8,640 five minute periods a month).
We then take the top five percent of these samples (5% of all the periods, or 432 five minute periods, or 36 hours) and throw them out because as seen in the diagram above, those are probably spikes that aren’t really representative of your typical bandwidth usage.
Your bandwidth ratio is determined by the highest remaining sample. This method of billing provides you with a number of advantages. Any usage bursts that are atypical of your bandwidth requirements are not held against you, and this equates to getting your atypically highest 36 hours of bandwidth usage free each month.
Here are some examples. ISP’s must charge for bandwidth by one of three means:
Thanks to http://www.seanadams.com/ (Site may be offline.—Ed.) for this discussion.
Your base cost before bandwidth includes hosting your streaming media on several sets of media servers distributed among multiple data centers on multiple backbones, centralized media distribution management, and gigabytes of storage on redundant network attached storage. The data center locations were not selected by cost, but by the appropriateness of their bandwidth to streaming. See the video delivery network page for more details about the network.
No, we do not charge differently based on the format. We charge based on volume, and we charge for certain paid media features or content protection features.
A one year contract gives you the best price, while on a month-to-month basis, since the main component of the price is bandwidth, price breaks are associated with buying a certain volume, just like a cell phone plan.
Please visit our documentation page to download “one sheet” marketing brochures and detailed documentation PDFs. We don't offer a lot of printed marketing materials because we prefer to spend our resources on building the technology to support higher quality streaming and making the network work more easily for our customers.
If you have any questions that are not answered by this web site, please don’t hesitate to get in touch with us. We’d appreciate your input, and we’ll make sure that we answer those questions for future customers in upcoming revisions to this web site.
We can put a new “on demand” or “live” streaming video client online, worldwide, as quickly as you can choose a video delivery plan and sign up.