Storage Switzerland Blog

Cable Survival Part III - Labeling and Naming

Tue, 16 Dec 2008 13:17:28 -0600

In the last two blogs we discussed the need for proper cable management, whether you are deploying a new data center or cleaning up an existing one. In this posting, we will continue in that same vein and discuss the need for the proper labeling of the cable plant for servers, network devices and storage devices. Labeling is just as important as properly strung glass, copper and power. Labeling for the cables is only the beginning,

Cable Survival Techniques Part II

Tue, 25 Nov 2008 15:51:19 -0600

In my last entry we discussed, what to do when you start fresh with cabling, now let’s discuss a cable plant you may have inherited with your new position, promotion, so on. This is most likely to be the situation you are in today if you are facing these issues. If you are walking into a disaster, unfortunately, there really isn’t an easy way out of the situation. It really depends on how bad the cable plant design and implementation was from the beginning. I have worked with and assisted many customers navigating their way out of this tangled mess. Some of the ways in which I’ve helped to get them whole again include:

Total Replacement – Process of pre-positioning the new cabling infrastructure inside the location to where they all need to be connected. Taking an outage, ripping down the old, and swapping out to the new.
Partial or Re-Use – Process of a disconnect and a reconnect of the existing cabling and movement into new cable trays, rainfalls, server cable arms etc.
Both of these processes can be deployed a bit at a time. If you can plan the event out over time pick a rack, or group of racks to attack, this minimizes the number of potential planned outages.

As you can see, this is really not an attractive way for any data center manager to perform these types of tasks. But when you have a mess to clean up there are always going to be some sacrifices. I have had real success planning these clean-up sessions, around data center outages, which most corporations have from time to time for patching or power upgrades, etc.

The real drivers behind designing, implementing and maintaining proper cable management from the beginning is the ability to provide adequate cooling to the devices (versus trying to push the air thru a huge mess of cables), but even more valuable is that it will greatly minimize the pain of troubleshooting when a problem directs you to physical layer. Believe me, as I have been in this situation at 1:00AM, it’s much less frustrating to be working on an issue when you can quickly and easily identify the cable infrastructure versus crawling around or under the raised floor trying to determine what is connected to what. A good cable plant will also prevent people from inadvertently pulling out the wrong cable and potentially causing a drop in application or network service. For example, if you were crawling under the floor into a spaghetti mess, there is always the possibility you may catch your foot, hand, screwdriver or whatever, on a power/network cable and cause a real outage.

If this is something you haven’t had much focus on because of time or staff constraints, I encourage you to do so, as it will ensure longer life of your equipment and your data center cable plant. This makes for a much happier sys-admin staff as well, at least in the long run, after the problems have been corrected.

One last point: assume you get to that utopia like state in your data center, it does no good if processes and procedures aren’t put in place to ensure that the environment is not only maintained but documented as well.

Once you have crawled out of your cable plant nightmares and come out clean on the other side, you will have a cable plant and data center that you can be proud of. It’s definitely not a sexy project to undertake but your sys-admins, operations and engineers will thank you for the time and effort you spent designing or cleaning up your cabling environment. Because at the end of the day, properly strung glass or copper will ensure you have as an efficient as possible environment.

Standards - Part II

Mon, 17 Nov 2008 15:39:45 -0600

In my last Entry I described in detail why manufacturers are so loathe implementing standards for anything except the bare bone physical components. Standards and profits are mutually exclusive. Make no bones, if everything was standardized all the companies that we love and hate would disappear tomorrow.

But something is amiss; there is another guilty party here. Let’s not forget that we the users are part of the problem. User’s get together and all lament the lack of standards. For example I cannot tell you how many times I have been asked as a sales person for various big companies, why can’t I order a standard Seagate disk for $X on-line from a myriad of sources and plug into my array? Why do I have to buy the same disk for 3 x $X from the manufacturer. My standard answer was, we test 100% the drives (and return the bad ones to the manufacturer who sell them retail), we have software on the controller that talks with our software on the drive that makes sure you have high data reliability, look ahead algorithms that reduce the seek and latency, etc. but I digress.

Let’s imagine you are operating a shop running a high powered database doing thousands of credit card transactions per second from all over the world. Your shop is adamant that no proprietary solutions will be implemented because you are going to be an open shop. All is working well, you have a heterogeneous shop and you have a staff that keeps everything working.

Now a salesperson that you know and trusts shows up with a solution that is 20 times faster, costs half of your current solution with no maintenance of software fees for 7 years, uses one third the power, technology refresh clause over the seven years with no fork lift upgrade, and half the footprint. By the way this is a proprietary solution. Yep, the march for a standardized environment has just gone out the window. Your next question is ‘Can I get this installed next weekend’?

The reason is that both the manufacturer and the end users are both looking for competitive advantages. The end user is thinking about reducing cost (remember Wall Street, and budget reductions), reducing risk, reducing staff by doing the same number of transactions with fewer staff, or increasing productivity with the same staff. Very compelling reasons for making an exception for this vendor. The manufacturer now has a strong selling point for charging a huge premium for the product. For end users the request for standards go out the window when the pressure of the business, the pressure from upper management to do more with less and the desire to be competitive takes control.

For the vendors the margins come from doing what no one else is doing and building a case that your solutions will reduce risk, increase productivity and add to the bottom line. For the end users the bottom line drives everything, finding a way to save even a penny on a millions transactions a second eventually leads to some serious money, and all to the bottom line.

Conclusion: I am all in favor of standards for the common parts. If standards were all there were to a selection we would all be driving the same car. Don’t get me wrong I’m glad for the transmission being P-R-N-D-D1-D2 in every rental car, but when I buy something for my business I want the best bang for the buck and many times these are proprietary solutions and not standards.

Climbing out of the Spaghetti and Coming out Clean on the Other Side

Thu, 6 Nov 2008 09:07:22 -0600

As a prior end-user one thing that always irritated me is none of the vendors seemed to have any real answers around cable and label management, at least not something you didn’t have to pay for a professional services engagement to accomplish. It was something I had to develop primarily on my own based on what worked for my company and the data centers I was working in. Someone once told me “anything worth doing is worth doing well” and this holds just as true for the cabling aspect of any data center.

Over the last few years I have worked in many data centers of all shapes and sizes and the one thing I’ve noticed is that people tend to focus on the state of the equipment, servers and storage more than the physical layers of the facility itself. This, in my mind, is even more important than any of those servers or storage. For example if you build a house and use cheap cement but buy really expensive furniture and the house falls in, there was probably a huge error in your project plan. I realize this may be a tad overstated, however I just want to convey to you the need to have an efficient, well-managed data center.

First off let’s discuss cable management and the need to deploy it properly from day one inception of your data center. There are many ways to tackle this process. Let’s look at a new facility first, being the perfect of most situations, as it allows you to start with a clean slate. There are a handful of categories for cable management which include:

Server Cables (Copper, Glass, Power)
Storage Cables (Copper, Glass, Power)
Network Layer (Copper, Glass, Power)

Starting with these, the first thing one needs to think about is how these cables should be dressed and/or tied down. When designing a new facility this is your one, or at the least easiest time to make some of these design decisions:

Cables to be run overhead or under the floor(Typically the power and the network cabling should be separated from one another if possible)
Usage of structured copper and glass cabling plants to punch-down panels, fibre tubs, raceways, conduit and retractable trays for splice management
Patch panels in the server or storage racks with structured cabling running back to the main network line-up
Cable management trays for the network devices
Cable management side panels to cover extra cable slack and tie offs
Rainfalls for the mass cable drop-in’s into the network row

As you can probably tell there are a lot of decisions to be made in order to provide your company with an efficient data center cable plant. Don’t fret, there are many places to go for assistance when doing your design, I am just pointing out some of the areas you will want to consider.

Now let’s assume you are starting to place your servers, storage and network gear into the data center. There are some things to keep in mind here as well. With servers, the first thing NOT to do is throw away the swing arm; use it for what its intended for: “Cables”. On storage if it comes pre-racked then the cables should already be dressed, other than the ones that connect to your SAN, ensure that those are brought in an advantageous way ensuring that you have free access and can tell where each one goes. On the network side, which will probably be one of the most cumbersome areas for proper cable management, ensure that there are ample cable trays/rainfalls provided for the significant amount of cables that will come into this area.

In my next entry we will discuss what to do if you inherit a cable plant.

Standards - Why the Pushback

Tue, 28 Oct 2008 05:54:38 -0500

I love standards; light bulb sockets, power cord plugs, street light colors, driving on the left (here in the states anyway), food quality, pharmaceutical quality, data processing…whoa, wait a minute. Standards for data processing; you mean like file structures, header files, caching algorithms; migration, error checking, encryption, backup and such?

I see, you mean CAT connectors, SCSI connectors, cabinet widths, HDD packaging, chip connectors and all the other mechanical parts that go together. But what about migrations between vendors, writing to another vendors array, moving a tape drive from one vendor’s library to another vendor’s library, taking a hard drive out of one vendors array and putting into another vendor’ array, backing up with Legato (EMC) and restoring with Veritas (Symantec) and let’s not even start on encryption and keys.

Does anyone see the problem with lack of ‘standards’ that really do something for the user? Let me state this flat out…standards will never take hold until whatever is being standardized has no added value. Blunt? Cold? Not in touch with reality? I know I have been hearing that for the last 20 years, back when I had to draw RAID pictures on a white board. The question that no one has been able to answer is: Why is one manufacturer going to do something to make life easy for a competitor? Why would EMC make migration easy for HDS? Why would HDS make migration easy for IBM? And I am sure IBM is working on a migration tool to move to EMC. Now picture the second tier providers; I am sure their engineering staffs are working on the same standards for their products. In Las Vegas we call this a sucker bet.

Standards carry no added value. Standards, when used, are what I call hygiene factors. A hygiene factor is something you have to implement, but the implementation does not allow you to charge more or give you an advantage in the selling cycle, but the absence can cost you business.

Here is an example of a hygiene factor; bathrooms. Finding out that a company has a bathroom where you are interviewing would not get you excited about taking a position. Now imagine interviewing for a job and finding out that there are no bathrooms, none, nada, zip on site. Would you take the job? Probably not.

In the IT world we see hygiene factors all over products; widths, heights, 110/220V, CAT/USB/SCSI connectors, KVM connections and so on. Do any of these factors cause you to write a PO? No. But have someone introduce a machine that runs off 330V, 600HZ and 5 phase power and all of a sudden we have a big problem; unless the benefits far outweigh the disruption, cost and impact, the product is not going to sell.

Standards are the fodder for organizations that look at the disparity of solutions and try to get manufacturers to agree to doing something for the end user that makes their life easier.

The problem with standards is that standards and profits are mutually exclusive goals. If every one produces the same product with the same features that work the same and is reliable; where is the ‘added value’ that cause the price to go up increasing margins and allowing wall street to be happy. There’s a true oxymoron ‘happy wall street’; but that’s another article.

The problem with standards is like discussing prison and dumps; everyone says we need them - but not in my backyard. Similarly the manufacturers all smile and nod politely and say that standards are good but ‘my’ product is so unique that the converting to a standard doesn’t apply (not in my yard).

But alas, the manufacturers are not the sole protector of proprietary solutions; there is an equally guilty party that bears far more of the blame for the lack of standards.

I’ll disclose the guilty party in the next discussion.