Three More CDNs Launch, Market Too Crowded

Amazingly, the number of new content delivery providers in the market continues to climb with three new CDNs launching in the past few weeks. The new entrants, which I will cover next week are Jittr Networks, SimpleCDN and EdgeStream. I think it's great that more companies are offering services in the market and that investors seem to have no qualms in pumping more cash into the industry. But we're now looking at over 40 content delivery networks and it's just too many. There is not enough business out there today to support so many providers, all offering different variations of the same service. On Monday I will be updating the list of CDNs that I track in the industry at www.cdnlist.com

I think choice is great and why not have as many choices as possible for any product or service? The problem being, in the long run, many of the CDNs are not going to be able to grow their revenue to meet their investors expectations. I keep hearing almost everyone say how they are going to give Akamai or Limelight a run for their money, but nearly all of the new CDNs, or those who have been around for a year or so, will do at most, 5-7% of Limelight's projected 2008 revenue. So far, only Level 3 is showing any signs of really growing their CDN revenue, based on the data they gave out last week during their earnings call.

That's not to say that all CDNs are trying to go after Limelight or Akamai or even want to become that big. A rare few of the CDNs make it clear that they don't want to be the size of Limelight and if they do $15 million this year they will be happy. Kudos to them for not giving into the market pressure of a new company thinking they have to launch to the market saying how they are going to take down the number two provider. What's wrong with being a smaller, profitable company not in the top three based on revenue? Nothing. Better you set expectations properly, your own and your investors, and survive for years to come in the market.

I hate to say it, but we're going to see a lot of cracks in the CDN sector starting 18 months from now. The market simply can't sustain so many vendors. If the market size was five or ten times what it is today, then all of these providers would have a shot, but it's not that big and won't be that big 18 months from now. For all the new CDNs, none of them seem to really have any idea what percentage of the market they think they can grab. While many of them say how they can take business from other providers, rarely do they say what percentage of the market they think they can get. I also don't hear from any of them what they think the market size is for CDN services in the U.S.

I'm all for new players in the market, giving everyone a fair chance, providing customers with more options and having the industry grow as a whole. But when you have so many providers in the space, all saying the same thing; we are cheaper than Akamai and/or our delivery is better quality, especially for high-bitrate video, then you can't expect to grow your business for the long run. And with more CDNs in stealth mode still waiting to launch, and telcos like AT&T and others taking more of an interest in getting into the CDN market, the number of providers for CDN services is going to take a real hit when the VC money starts to run out.

Comments

[Jack]

SimpleCDN is far from being a true CDN - I would recommend you do some more research on what they really are before lumping them in with the other CDN players.

As for being crowded - the majority of the new entrants seem to be P2P players where the CapEx requirements are much lighter to get started. Look for consolidation in this sector - long overdue IMO.

[Dan Rayburn]

"true cdn", what does that mean? The term CDN is as generic as you can get these days. Most of the new entrants are NOT P2P based like EdgeCast, EdgeStream, Voxel.net, BitGravity, Highwinds, etc...

[HankG]

Right, there is no definition of a CDN. One would say they need a lot of edge servers. Another one would say they need a distributed network like P2P. Another one would say that a high performance network with a central server farm is also a CDN.

My definition of a CDN is nothing more than a optimized hosting and delivery service.

IMHO CDN's with many edge servers are becoming too expensive to rollout and manage and P2P networks just don't offer the quality experience (especially not for realtime delivery). My bet is on network oriented CDN's. Fast pipes, cheap connectivity and massive media server farms on a limited number of sites.

[Steve Lerner]

We define a CDN as an outsourced content delivery service with the following attributes:

- multinetwork
- multiprotocol (http, https, RTSP, MMS)
- multifunction (small object caching, large file download, SSL, web page acceleration)
- internationally deployed
- optimized for performance

By this definition, fewer than 7 companies were very truly CDNS- with the rest being variations on managed hosting services.

Managed hosting does a fine job for many needs, but it can't address the needs of a customer who needs a CDN as defined above.

[Frank Wilson]

Why is it that SimpleCDN comes with flat-rate pricing, and then everyone else in the industry lines up to take their shots? Please explain why SimpleCDN is not a "true cdn". :-)

As a true HTTP file delivery service, SimpleCDN currently delivers files via HTTP from multiple locations. Doesn't Akamai do that? Direct links to files, download accelerator support... seems the same to me.

SimpleCDN also converts any uploaded video file into multiple VP6, H.264 and VC-1 video files for progressive download, as well as RTMP and MMS/HTTP streaming delivery to Flash and Silverlight. Akamai streams videos too, right?

I believe that HankG said it best... "IMHO CDN's with many edge servers are becoming too expensive to rollout and manage and P2P networks just don't offer the quality experience (especially not for realtime delivery)."

SimpleCDN believes it can provide "Akamai Like" services for lower prices, with fewer hosting points and intelligent network design.

Even with the telcos filtering P2P traffic, I would still argue that peer-assisted delivery solutions are still "Content Delivery Networks".

The rest of 2008 and 2009 should be very interesting. Many of the new "HD video" CDN's have their own player, like Move Networks and EdgeStream (or other peer assisted networks). Will these "software needed" CDNs really start to catch-on, or will people/companies stick with CDNs that stream directly to Flash and Silverlight?

[HankG]

Steve, there is no big performance / redundancy difference in a CDN that has many edge servers and a CDN that has one mega core server farm and a great international network. IMHO, anyone who can offer high capacity and high performance media delivery (whether it's through networking or edge servers) is a CDN. And file delivery isn't that interesting. It's all about the streams.

But there is a big investment difference. I know some centralized CDN's like StreamZilla who have been profitable from the start and I know many edge CDN's like Limelight who will be in the reds for years.

I don't believe in CDN's that rollout their own client software. Customers just don't install custom players. Stick to the standards.

And I also don't believe in Peer2Peer or Peer Assisted services. Especially not for streaming or realtime delivery. Even Joost uses servers for 75% of their VOD. And 100% for their live streams. So much for peer assisted delivery...

P2P is such a network hog. Look at the UK market, where the BBC streams TCO was 3 million per year. They switched to P2P which equals an annual TCO of 1 billion (333 times as much!!!) for the access providers... with the same audience, but crappier video.

[hankG]

Frank, EasyCDN was NOT the first to come with flat-fee delivery!
See www.streamzilla.eu/dedicated.html (yes I am biased).

Flat-fee streaming from dedicated media servers, since 2003.
Server, pipe, software, SLA, unmetered traffic. Fixed monthly fee. HTTP PDL, RTSP, MMS, RTMP and FLV semi-streaming similar to YouTube. A+ International performance.

You will pay less than €10 per Mbps, TCO. Recalculated price per GB can be less than € 0,03, TCO.

[grinsandfun]

My question is SIMPLE - when are you going to stop your US centric focus only? There are groups (as you read between the lines of your posts) that you seem to have built a bias against and also from reading the shunt (above) that goes against the initial PR for CDNetworks - US launch (which did state they are targeting #2) - why don't you acknowledge that there are Asian CDN's (including ChinaCache - which I see also having a US presence now also) & European CDN's that could make a considerable impact against the current #1's and #2's.... especially Limelight - since they have traditionally sold on price!

BTW - I also listened to LVLT CC & found it amusing that they (if you listened to how they positioned their Content Markets Unit), could fool an investor into thinking they did $100 MM in Content Market sales for the quarter - the foolish one would think - WOW - their CDN grew to being #2 in the CDN revenue - not the case - they lump in sales to CDN vendors also into this fold - not just CDN product.

Looking forward to your "Global Perspective" :)

[Steve Lerner]

In regards to:

"Steve, there is no big performance / redundancy difference in a CDN that has many edge servers and a CDN that has one mega core server farm and a great international network."

Do you have any case studies to prove the veracity of this? I'd love to see some data.

On the other side of the argument, the veteran CDNs can show many examples of what happens when a performance based CDN is load balanced against a 'single well networked location' and the bandwidth drains off the single origin and distributes to the CDN edges.

During my years at Speedera we'd demonstrate this all the time to our customers- a performance optimized CDN trounces any single location if *performance* is the load balancing criteria.

Feel free to send all questions to steve@ramprate.com

[Frank Wilson]

Hank... That is SimpleCDN, not EasyCDN. ;-)

I think you are missing the point of SimpleCDN. First, the plan you mentioned from StreamZilla is 9,950 Euros, or $15,328 dollars per month. Second, you could only achieve a net cost of 0.03€ per GB if you had perfect 100% utilization around the clock. We all know that is not going to happen, so your effective cost per GB is much higher.

So if I have 100 training videos (assuming each video is 900 MB), you are looking at $3,500 for SimpleCDN for a year (or even 5 years) for unlimited traffic and bandwidth, versus $183,936 for a year at StreamZilla (or $919,680 after 5 years), and you could only ever push out 1,000 Mbit/sec across all of your videos.

I don't want this to sound like an advertisement, and I am sorry if it does, but the point I am making is that unmetered and unlimited delivery plans ARE NOT the same as flat-rate per file pricing. I hope that you see the distinction. I believe SimpleCDN is the first to offer pay once, stream & deliver forever pricing.

[HankG]

Much has happened in the past years which will change how people look at CDN's.
5 Years ago, the backbones were limited and expensive. So Edge servers in local regions were cheaper.

CDN's actually do not optimize for performance but optimize for the ideal 95 percentile load of their Edge capacity. CDN's still redirect users to edges in a remote region to manage their costs. This still happens and it is bad for performance.

Today, there is extreme carrier backbone capacity at low rates. And don't forget Internet Exchanges. It is more expensive to rollout and manage edge servers across the globe than to use these backbones and exchanges.

You may say that latency can add up. It does. A bit. Some remote regions can suffer from this. It is not economical to invest in many Edge servers in remote regions with a market value of less than 1%. And it may be harder to guarantee more than 25Gbps capacity in some regions. But the number of customers who demand such capacity is small.
95% of all the customers can be served with a central media server farm and a great international network. At perhaps 20% of the investment. Most CDN's are fighting and wasting investors money to get the remaining 5% of the market.

But today it is not carrier backbone latency that is the issue: the true limiting factor today is the ISP network capacity. You can't solve this by having edge servers within their networks. You also can't solve this by having private peers with the ISP's.

IMHO, it's no-Edge or micro-Edge. Micro-Edge is where small edge servers are placed throughout ISP networks. Each ISP itself is a CDN so their own backbones are offloaded. We do this a lot for ISP's. And then you can link the ISP's to build larger CDN's. That is where the real benefit is.

What you absolutely should NOT do is use Peer2Peer to deliver your content. It may look like a cheaper alternative for the broadcaster, but effectively it moves all the costs to the ISP's. It actually increases the costs. BBC wanted to save 5 million dollars on streaming, switched to P2P and this will cost the UK ISP's 1.5 billion dollars to upgrade their network!!! Penny wise... pound foolish. ISP's are massively blocking P2P for this, bringing down the (already bad) performance.

About the Flat-Fee: 100% is indeed the maximum load. Most customers can optimize to 67% which means a recalculated price of $0,04 per GB which is still extremely low. Guaranteed capacity (non overbooked, not shared capacity on a CDN), great international performance. Stream and deliver in any format. The cap can be 100Mbps, but can also be 1Gbps, 5Gbps or 10Gbps, etcetera, whatever you want. You can grow along with the volume. Plus there is a OverFlow option which allows you to burst temporarily above your cap.

I don't understand your training video calculation. First I think that pay per file has no match to usage. What if the file is never watched? Or what if the file is watched 100x per second? Does your business case still live?

An average training video is watched less than 100x per day. If it's encoded at 1Mbps and lasts 10 minutes (which is way above the average online viewing time), = 100 * 40MB traffic per day * 31 days = 121GB traffic per month. Will set you back less than $300 per month / 3,600 per year.

Suppose that the training videos are popular. 100 simultaneous views, average viewing time 15 minutes, day in day out, at 1Mbps each. 300,000 views per month. This will require 100Mbps flat-fee. $ 3,000 per month / $ 43,200 per year.

[Steve Lerner]

The declarative statements about edge servers are all based on assumptions with no data or reference behind them.

Actual usage will always show the truth... and even if you had the data, your assumptions are based on video usage and video usage is not the key value-add of a CDN- the value add is an overarching set of performance, reliability, and analytics across all types of content.

Try a central server for that type of business, and you'll see what happens... its been tried before. Notice who is able to stay around and be profitable in the CDN business, and who isn't.

The difference between success and failure, if you look at financial statements and QOS tests, always comes down to performance.

Talk to anyone who has run a large scale load balanced CDCN and you'll see.

-Steve

[Frank Wilson]

Hank,

You are doing a disservice to the community here with posts like that. It sounds like you work for StreamZilla, saying that CDN capacity is overbooked and shared, but then you say "We do this a lot for ISP's." [regarding micro-edge servers on their networks], so maybe you work for a CDN. I don't know! ;-)

First of all, you are trying to compare our CDN services to a dedicated media server in a single location. But we'll look past that for now...

Let me help you with your calculations...

We are going to take your example to start. 100 training videos watched only 100 times per day. They are 1Mbit/sec encoded and are 10 minutes long. Okay here we go...

10 minutes = 600 seconds.

1Mbit per second * 600 seconds = 600Mbits per video.

600Mbits = 75 Megabytes per video.

75 Megabytes * 100 watches = 7500 Megabytes.

So since each video is watched 100 times per day, and is 75 Megabytes, that means that each video is going to use 7,500 Megabytes of bandwidth per day, or 7.3 GB per day. We have 100 videos, so that means 100 * 7.3 GB = 730 GB per day of transfer.

Okay so now we have 730 GB per day of bandwidth usage. Now we have to work backwards a bit to find the correct "StreamZilla" plan, since we do not know if all 100 views are going to occur at the same time, or if they will be spread out. Let's assume that the views are spread out over 12 hours, and you want a 2x ceiling to ensure your videos are delivered smoothly.

So here we go again...

730GB delivered in 12 hours.

That is 60.8 GB delivered per hour.

60.8 GB = 498,073 megabits delivered per hour.

Since there are 3600 seconds in an hour...

498,073 divided by 3600 = 138 Megabits per second of video delivery.

But remember we needed a 2x margin, to handle traffic spikes, etc.

So we really need 276 Mbits/sec of capacity. Okay so I am now on the StreamZilla website. Dedicated media servers. Wait a second, they have a 250Mbit and a 500Mbit plan. Well what should we do? Lets give you the benefit of the doubt, and only go with the 250Mbit plan. Lets just hope that the views really are all spread out, and not more than a few people try to watch each video at the same time... otherwise "buffering" "buffering" "buffering"

So we have the StreamZilla "Pro Media Server".

250Mbps bandwidth and 750GB of storage. Seems like this should work.

Well look at that Hank. It costs 3,950 EUROS per month.

That is $6,098 dollars. $6,098! Per Month! Wait, if you say you are in the Netherlands instead of Europe, the price is only 2,495 EUROS per month, or $3,852 dollars per month. $3,852! Per Month!

Wow I am tired. Wait, let me use SimpleCDN instead.

Okay 100 videos, they are 75MB each. Looks like they will be 9 credits each, or $9 each with no bulk credit discount.

So $9 x 100 = $900.

$900. Done. And SimpleCDN will convert them and stream them in 2 Flash VP6 formats, 2 H.264 Flash formats, and 2 VC-1 Silverlight formats... Free!

So yes Hank, SimpleCDN IS the first to offer flat-rate pricing. Your service is $3,852 per month. At least. Mine is $900.

$900. Once!

- Frank

[Dan Rayburn]

As an update to this post, I would not recommend SimpleCDN.com - We have gotten a lot of feedback from customers who have used the service and have gotten very poor service and have had major problems with billing or getting refunds.