Internet of Things, Robots and Fully Automated Data Center

This is a light read for a weekend piece.

Many reports are saying how Internet of Things will generate so much data that more data centers are needed to store and process such huge amount of data. This post is not about that sort of impact. Instead, it is making sense of the amount of things aka sensors that are in a data center and whether we can now fully automate a data center.

No more than ten years ago we have bank teller, customer service agent, post office clerk, security guard, soldier, etc; nowadays it’s ATM, interactive helpline, self service stamp dispenser, two way Comm with camera (the guy maybe remote location), and a joystick welding jock that “fly” a UAV. We are dealing more and more with automated responses and people that are time-spliced to deal with those few that got through those pre-programmed decision rhombus.

All modern data centers are highly automated, chiller gets turned up and the CRAC switched from standby to run state when the heat load is sensed by sensors, and vice versa is done when heat load drops. CCTV records faithfully non stop 24×7, and door is released based on authorized card access.

Human activities for a data center, especially one that is at or near full capacity, takes place usually in reaction to events such as visit, incident, maintenance activity, and delivery of supplies. Pre-scheduling all of the non incident events makes flexible scheduling of staff an opt especially when such routine drains scarce experienced data center staff is not an efficient use of valued manpower.

If you watched the 2009 movie “Moon” by Duncan Jones, a single human operator monitor an entire automated mining facility with the help of GERTY which is the station artificial intelligence. By the way, I strongly believe a shift has got to have at least two duty personnel both to have separate do and check duties plus to check on each other well being during hours. Is it possible now to have that kind of automation in a data center?

Before Internet of Things get popular as a term, our data centers have been installed with hundreds or more sensors, temperature, humidity, heat, airflow, current ampere, circuit trip sensor, motion detector, electronics lock, automated door, water leakage etc. These sensors sends signal to the central management systems that personnel get pinged to react and manage any anomaly.

We assign staff to monitor maintenance specialist from the manufacturer appointed service company doing what the qualified service personnel should know very well to do and our staff strength limits assigning too many maintenance activities with enough capacity to respond to incident caused by anomaly.

What about those situations or places to faults which are not fully monitored?

This is when we should have a well documented and growing incident database that collects incidents from every data center that is out there. Situations such as the high pitch noise of chilled water pump shaft wobbling giving off pre-failure sign tells us that we can put camera and noise sensor to monitor the chilled water plant room.

The possibilities are there.

Going one step further, how about deploying robots? Looking at these latest development on robots, it is a possibility now or in the near future to see these in a data center monitoring maintenance work, doing routine check, patrolling, etc:

http://www.fastcoexist.com/3049708/meet-the-scary-little-security-robot-thats-patrolling-silicon-valley

http://money.cnn.com/2016/02/24/technology/google-robot/

Just to clarify, I am not advocating to replace human with robot, but to take away menial and routine task that are more efficient if monitored more regularly.

The case of the wobbly cam shaft of chilled water pump is a real incident at a data center site. There is another case where a corroded chilled water pipe that burst and leaked water which got onto the UPS at the floor below. These places are patrolled by 24×7 duty facility engineers who patrolled at 4 hours internal, but the previous round has walked the place an hour ago and the incident took place.

People are still needed, like the operator in the “Moon” movie, but the operator deal with the anomalies and gets thing moving along. We want our staff to be challenged and do the tasks to keep our data center humming along nicely.

So, the Internet of Things can be more fully deployed to plug the gaps where we still rely on human patrol and check to detect issues, or put a staff to monitor maintenance specialist from the manufacturer appointed service company doing what they should know very well, it may already happened in a third party data center service provider whereby site are monitored remotely during off visit hours . Going further to see lots of fully automated data center will be a reality in the next 10 years.

Talks of a few fully automated data center:
http://www.datacenterknowledge.com/archives/2013/05/22/the-data-center-of-tomorrow-totally-lights-out-within-5-years/

http://www.datacenterknowledge.com/archives/2011/10/11/aol-launches-new-lights-out-data-center/

Internet of Things, Robots and Fully Automated Data Center

What are the public listed Data Center companies or REITs?

Published on 18th February 2016.

Recently, I thought of the question of what are the public listed data center companies or REITS? I put my favorite search engines to work and a bit of checking about, what I found is interesting to say the least.

While the public don’t hear much about data centers, they are here and nowadays there are quite a fair number of public listed companies with significant portfolio of data center assets or are highly dependent on data center for a significant share of their revenue.

Now, for the purpose of this post besides it must be listed on a public stock exchange, I define the criteria of what qualify as a data center service provider of having met at least two of the following conditions:

  1. It has more than 6 data centers under its control.
  2. The data center business account for more than 25% of its revenue
  3. Contribute to at least 10% of its profit

Listed data center REITs definitely fits the above profile. How about Telcos? Telcos do have a large number of data center, for example the big three telcos in China owns 75-85% of the data center market between them, but data center is not significant enough for these big telcos, so while you can buy China Telecom, China Unicom, and China Mobile shares, let’s put them aside for the purpose of this article.

I cast the net wide, not limited to the US stock exchanges, and the list I had come up with is as follows (s/n, name, stock symbol or numbers, stock exchange) :

  1. Digital Reality (DLR), NASDAQ
  2. Dupont Fabros Tech (DFT), NYSE
  3. CoreSite (COR), NYSE
  4. Quality Technology Services (QTS), NYSE
  5. CyrusONE (CONE), NASDAQ
  6. IronMountain (IRM), NYSE
  7. Equinix (EQIX), NASDAQ
  8. InterNAP (INAP), NASDAQ
  9. InterXion (INXN), NYSE
  10. IOMart (IOM), London
  11. Dada Group (DA), Milan
  12. Retelit (LIT), Milan
  13. Rackspace (RAX), NYSE
  14. NEXTDC (NXT), ASX
  15. Vocus Communications (VOC), ASX
  16. Asia Pacific Data Centre (AJD), ASX
  17. Internet Thailand Plc (inet), BKK
  18. CSF Asia, London AIM
  19. 21vianet (VNET), NASDAQ
  20. DrPeng (600804), Shanghai
  21. Keppel DC REIT (AJBU), SGX
  22. Grand MING  group – itech data centers (1271), HKEx
  23. Sunevision – iadvantage (8008), HKEx

Note:There would be more but some were merged under M&A (e.g. Telecity by Equinix) or are bought and turned private (e.g. COLT by Fidelity).

It turns out that there are whole lot of data center service providers we can look at on the stock market, read their reports and consider to buy them, and in Asia (including Australia) there are at least 10 listed entities. As suggested below in the AsiaOne link, the number of such listed data center REITs or service providers will grow as these companies tap the capital market to grow their reach and grow with their customers, and achieve better economy of scale.

There are a lot more data centers that are privately owned by such as Switch/SuperNAP, Global Switch, SunGard, or H5.  And there are big data centers owned and operated by those huge Internet companies like Google, Microsoft, Apple, Facebook, eBay, Alibaba, Amazon etc.

Another piece of interesting information, there are a bunch of other developers or REITs that happens to have one or more data center property under its belt. E.g. in Singapore there are Kingsland, Ascendas REIT, Mapletree industrial trust etc.

One option to consider is to invest into listed companies that its main line of business is also growing and has sizable DC portfolio, KDDi with Telehouse.

Some private data center companies bonds on the stock exchanges, for example DigiPlex and Global Switch had issued bonds via Oslo and London stock exchanges respectively.

Finally, these data centers serves big clients such as Amazon, Google, Microsoft, Alibaba, Tencent, so you may want to consider buying the big user of big data center spaces since this indicates that they are growing strong.

 

Reference:

What are the public listed Data Center companies or REITs?

The Singapore Data Center supply in 2016, rosy or gloom for next three years?

singaporeDCmapQ1-2016

Published on 23rd February 2016.

Is Singapore Data Center Market alright?

A February 2016 Straits Times article (reference 1) titled “Data Centres Shine Amid Property Gloom”, paints a picture that the data center industry will do better than the property sector and even the economy as a whole. I am concerned about the positive tone of the article given that the Singapore economy is predicted to be on recession to a flat year in 2016 (reference 2).

The support for the article is via two main premises:

  1. The article cited that there are rising demand from finance and tech companies.
  2. Government support to develop data center business in Singapore

 

Optimistic Demand

The first premise is that the demand will chew up the supply and things will be all well by early 2018.

Let us look at the second premise which is easier to deal with. The government support to develop data center business will only mean new player or existing player can more easily develop new data center facility, i.e. it deals with the supply side of the demand-supply equation. If the government can target the data center industry to give incentives or rebates to lower cost, it will make Singapore draw more data center clients vis-a-vis Hong Kong and the rest of South East Asia. But, it is not the norm for Singapore government to do so and so far it has mainly facilitated via designating and promoting the Data Centre Park.

I will be more than happy if the government attempt to influence the data center market by acting on the demand side directly or indirectly but it is not the case here in Singapore although our economic development agencies (EDB, IESingapore, IDA, etc) has and continue to promote and draw companies to Singapore to set up and grow their business.

Back to the first premise, it is quite counter intuitive given the examples in the article of tech companies Microsoft and Google. The fact that these tech giants are building new data centers only meant that they are building for themselves and not taking up the new supply that are coming up.

The example of OCBC also cited by the article did not help the case. OCBC bank is building its own data center facility and will relocate its existing data center to its own new facility? Therefore no OCBC take up of third party data center space. Increase in data center space demand? Nil. Coupled with the news that the financial service industry has seen layoffs since Q3 2015 until recently with the Barclay layoffs of 1,000 staff in Asia (reference 3), it is more likely that banks will review their data center demand and will more likely delay or reduce the pace of new space take up.

In one of my earlier post, I shared on surveys for data center demand, there is likely to be an over-counting of demand that needs to be moderated and I suggested divide by a factor of two.

The data center market is a connected market and not an isolated one. Singapore enjoys being the network hub for South East Asia (“ASEAN”), and it has half of the ASEAN data center market. Still, Singapore’s 5 million people is very small compared to the 600 million people in this region and investment can and has sometimes flowed directly to the other countries. For example, NTT has bought CyberCSF in Jakarta and had announced doubling the capacity of that data center.

At this point, I think it is clear to the reader that I think that the demand for new data center space needs to be carefully examined and considered for the data center industry stakeholders.

Let’s Count All The Supply

I see two main issues with the supply. They all come to a sum that is bigger than cited.

First, the supply is underestimated. i.e. same site demand that can be met within the same facility. Don’t get me wrong. New capacity of 47% coming up all in one year is huge for the small city state of Singapore! I think it is never heard off in a mature data center hub for a long time. It is highly likely there are expansion projects here and there within the existing data centers to take up some of the same-site demand. Clients are sticky (to a certain extent, I will examine and share my view via a future post) .
1. ST Elect (AMK)
2. 1-Net North
3. DRT Loyang
4. Singtel West
5. STTelemedia/Starhub Mediahub DC
6. STTelemedia Defu DC
7. Telin (DC Park)
8. Keppel T20
9. Global Switch

What is most interesting by this point of the examination into the supply side is that there are a number of third party data center facilities that were opened in 2015 or earlier which are brand new and have certain level of capacity.
1. Kingsland DC
2. Equinix SG3
3. Telstra/Pacnet DC
4. IO data center
5. NTT Serangoon data center

I agree with the article that it is true that these new capacity and existing capacity will mean it will be a tenant dominated negotiation for their renewal or expansion, and they may take the opportunity to shift to a newer or more resilient facility, what is left vacant will be inventory that is harder to lease out.

Some Suggestions to Maintain or Enlarge your Pie of the Market

I have some suggestions for the data center service provider to consider:

  1. Be prudent. Consider delaying investment or build or splitting the capital investments for the data center infrastructure into more phases to be just one step ahead of demand rather than one lumpy implementation.
  2. Review your client retention strategy.
  3. Pursue same-site client expansion.
  4. Improve the appeal of the vacant, especially older facility, by consolidating and enhancing the data center infrastructure and space on a bigger scale.
  5. Do not ignore potential small wins, they add up and also reduce the risk profile of relying on a few major clients (who has bigger negotiation power when supply is more than demand).
  6. Enhance service offerings, differentiates, talk to clients to draw out what will make them buy more services or will make them happy to stay (better network connectivity, cloud exchange, etc)
  7. Look at how migration works and put in effort to understand potential client’s concern about moving to your site (e.g. reference 4)

Last but not least, hang on to your valued customers, maintain a healthy communications with all levels of the client.

Reference:

  1. http://www.straitstimes.com/business/property/data-centres-shine-amid-property-gloom
  2. http://statestimesreview.com/2016/01/24/ntuc-expect-31-more-retrenchment-this-quarter-as-singapore-goes-into-recession/
  3. http://www.straitstimes.com/business/banking/jitters-over-looming-job-cuts-in-banking
  4. https://www.cgi.com/sites/default/files/white-papers/cgi-state-and-local-data-center-consolidation-white-paper.pdf
The Singapore Data Center supply in 2016, rosy or gloom for next three years?