body’s wedged open the door between the two. Of course, hardware equipment tends to fail more often than buildings catch fire, but when we’re really designing for true redundancy, all factors need to be on the table. Total mirrored hardware systems in disparate, geographic locations are cost prohibitive, even for large broadcasters, but in the brave new world of integrated, CiaB (channel-in-a-box) and virtualized systems, it no longer has to be! How many “9s” do you want? Cloud providers tell you that you can have as many “9s” as you want - and can quickly tell you what that equates to in the number of minutes downtime you can expect in a year. Greater reli- ability, with dual or triple redundant configura- tions, all comes at a price however. In practice, I think a lot of broadcasters may be overstating how reliable they need things to be. When you dig down a bit further, what they’ve achieved themselves in the last five years might be impres- sive (the signal was OK leaving me!), but if they have a fire tomorrow or an aircraft lands on the building, it’s going to wipe them out. So, on one hand you could argue they’ve been lucky, (no air- planes) and they’ve got reliable equipment - but are they really comparing like-for-like? A cloud provider’s comprehensive redundancy is at a scale far beyond what can be created “in house”. Some cloud providers aren’t even talking about the number of nines anymore, as they can point to 100% uptime for some services, world- wide. A key “take home” point here, is that the all cloud providers say you need to design your system to take advantage of the inherent redundancy that the cloud can provide. So, the onus is on us, the application developers, to correctly utilize cloud APIs and services into our solutions and with it the ability to tap into their redundant infrastruc- ture. We must not fall into the trap of exactly replicating the designs and architectures we are used to. How many “9s” do I really need? Broadcasters have always over spec’d things in 38 • Broadcast Beat Magazine • www.broadcastbeat.com the name of seeking greater reliability. However, if you’re running a 24/7 broadcast service, rep- licating everything in the cloud and building-in vast amounts of over-capacity (in cloud-speak – overprovisioning) this might cancel out the very reasons you want put it in the cloud to begin with. Because it’s more flexible, the cloud allows you to design different levels of redundancy for differ- ent services. When we think in this way, you can easily imagine a scenario where you can dynami- cally over-provision during critical broadcasting periods, such as sports finals, etc. This would allow you to guarantee a higher level of protec- tion during a Super Bowl or playoffs week, which you could then scale back for the reminder of the season. Taking a 30 second outage hit during a Super Bowl commercial is a much harder financial hit than the same outage during a late-night real- ity show. In this way, business rules logic can be applied to automatically scale redundancy in line with the agreed upon revenue risk. When designing for redundancy in the cloud, it’s important to ask for something sensible. What do you really need to duplicate? Making fair com- parisons with traditional hardware systems is not always straightforward. If you elect not to make the leap into the cloud - which still makes a great deal of sense for many scenarios - you still have to put your equipment somewhere, cool it, power it, maintain it, account for hardware refresh, perform routine compliance testing, etc., all whilst maintaining competent staff. You can’t just look at the initial equipment spend that you would have made on the project, you have to factor in all the things you are no longer paying for. Most importantly, you have to draw comparisons on scalability and think about those airplanes. Ian is one of the founding directors of Pebble Beach Systems, having worked on software projects in the broadcast industry for over 25 years. He has been instrumen- tal in the development and implementation of the com- pany’s range of automation and channel-in-a-box products at broadcasters and service providers around the world. Ian has also worked as a consultant for companies such as Reuters, NTL and The Walt Disney Company.