5

I'm looking for a term for use in documentation, and felt that SF users would have the more experienced set of terminologies compared to other sites.

At work there are development servers in the office (e.g. New York), development servers in datacenters (e.g. New Jersey, Connecticut), and UAT and production servers closer to clients (e.g. Toronto).

Among the office development servers and servers in certain datacenters, the mounted NFS partition for /home is shared—one and the same. Other servers however, even ones that might be physically located in the same datacenter as one of the aforementioned servers, mount a different NFS partition, so that my ~acheong is distinct between the two _____________ (groups? zones? networks?) At the UAT and production levels, the _____________ fragment more and more (understandably, as geographically the servers are more scattered, firewalls, etc.)

What is the word that best fills those blanks?

"Groups" seems too generic. "Zones" might be confused with the fact that we zone servers to create several virtual ones. "Networks" might be misleading because the servers aren't (always) on separate networks—they just don't always mount the same /home partition.


As I finished typing this question and considered what tags to use, I found , which, I kind of like, but I'm not sure if it sounds awkward to write, "This special .cshrc must remain synced between the two network shares."

1
  • Since you are asking about home directories, how is your user account info distributed, is there anything resembling an ldap tree spanning the whole org? Apr 12, 2014 at 8:54

8 Answers 8

11

I would tend to think of these as different environments.

5

"Subnet"

In a perfect would, we might refer to each grouping as a subnet, based on their network subnet mask, from which their interconnectivity is defined at the base of the network architecture [+1]. In this way all nodes would be relatable to their grouping based on their use of distinct private IP addresses under the appropriate geographically stratified subnets.

But this term is meaningless foreign jargon in our societies [-1].

"Silo"

The terms "silo" and "information silo" carry perjorative connotations, and that negative portrayal may be inappropriate for briefing less-technical audiences, especially senior management [-1].

Despite this, this term is especially well-suited to explaining the infinitely complex layers of a network architecture, because it offers an easy visual reference to farmland silos that stand tall, stiff and unopposed. This term therefore also subtly connotes a level of isolation or separation from other network elements, which in this case is quite accurate [+1].

"Environment"

While the term "environment" is frequently used to describe network architectures, I believe the usage more often indicates the entirety of an organization's interconnected networks, i.e. a secure, closed, private network would be considered a separate environment by virtue of it being closed from the other network [-1].

To add further complication, the term "environment" is the de facto go-to term for developers and programmers when describing programming environments often facilitated by virtual machines; the term is often cherrypicked into the frequent usage "dev[eloper] environment" [-1].

Nevertheless this term does accurately convey the intended meaning without inducing marked cognitive dissonance in our societies, especially in the usage "network environment" [+1].

Others, like "Cluster", "Array" and "Cloud"

The term "cluster" is notably disputed in the answers on this page for their technical accuracy in signifying the intended meanings, likely because of the prevalence of varied usages [-1].

This puts "cluster" in a similar position to "array", which since 1987 adopted the usage Redundant Array of Inexpensive Disks (RAID).

In brief, terms that are in frequent, strata-independent usage for specific purposes in other information technology domains, may pose a greater challenge to repurpose for this particular meaning.

Consider: "Layer"

An alternative term that could suit this purpose is "layer", especially in the usage "network layer", as it following the same network analogy as The Onion Router, known widely today as the anonymity network Tor [+1].

"Layer" could be seen to be particularly well-suited to this purpose because of it's wide acceptance relating to systems isolation [+1].

In order to incorporate the geographical distinction, some "layers" could be referred to as "geo[graphical] layers", while others might be "global layers" because they are not specifically stratified geographically.

However, it is arguable that "layer" connotes a higher degree of network isolation than the degree indicated in your question. For that reason, "layer" still may not be the best suited term for this purpose [-1].

The Real Question

Language is very fluid, dynamic and in a constant state of flux. English is a key example of that constant change, being heavily impacted by contact and coexistance with other languages. It's 2014. This industry is dominated by brands with names linking together phonemes that do not always comprise otherwise intelligible morphemes or that are unintelligible in their combination: twitter, Google, tumblr, instagram, feedly.

But that's just brands, the jargon comprising our IT geek speak could be seen to be even more esoterically derived, especially in common usages like "ethernet", which might otherwise be understood as a some sort of fishing net-like tool in an existential alternate universe.

Given the malleablity of language, there is little preventing you from innovating a semantically distinct, new term specifically for this purpose and then advocating it's usage. Consider the brands Saltstack, Ansible, Chef and Puppet: two seem centred in the kitchen and food, and the other two are centered in science fiction literature and 20th century geek speak.

You may be best off using innovative terms from the same origins. Consider: each node (end-user computer or server) a "chocolate cookie", all on various "cookie trays" (Organizational Units (OUs) in an Active Directory), with various types of "chocolate chips" applied to each "chocolate cookie" as they "bake" (push/pull/poll a server for Group Policy Objects) in various "ovens" (geographical groupings) some of which are processed in the same "master pastry factories" (datacentres), or "supplier warehouses" (outsourced cloud Infrastructure-as-a-Service).

This terminology may sound outlandish, but it's not that far off from the words of the Circus used by John le Carré in Tinker Tailor Solider Spy.

So the real question is: How innovative are you? versus How innovative are you allowed to be? versus How innovative ought you to be?

3
  • 3
    As delightfully thorough as this answer is from a {psycho,socio,}linguistic standpoint, it doesn't really answer the question, which is What do I call this so the other technologists don't look at me funny and ask what the heck I'm talking about? :)
    – voretaq7
    Apr 11, 2014 at 21:31
  • 3
    @GuyHughes You're wearing out my dictionary.
    – rtf
    Apr 11, 2014 at 21:51
  • 4
    @GuyHughes Sorry, but I think you're wrong here: Technical language requires a common vocabulary, and there are words for this already (like "Environment"). You're certainly welcome to your opinion, just as you're welcome to call anything whatever you want, but we'll have to agree to disagree on the practical value of creative naming here.
    – voretaq7
    Apr 11, 2014 at 21:58
4

Silo, or environment? The term "silo" is frequently used in Citrix XenApp environments to indicate a group of servers that serve the same app or set of apps.

2

I agree with squillman and mfinni 100% on calling these 'environments'.

Depending on your environment (yes, pun intended), maintaining a dedicated and format name is extremely useful. For example: 'Home XYZ Environment' and'Home ABC Environment'. Keeping these names specific to exactly what you are referencing and uniform is extremely handy.

More than once, I've come across cases where things aren't clearly differentiated. The same is true for the naming convention, both with people and within documentation.

1

I would tend to call them what they are - network shares, mapped drives/folders/directories, mounted drives/folders/directories, mount points, home directories - or something of that nature.

I think I'd favor "mount points" or "home directories," but wouldn't see a problem with any of the terms, as they're all accurate.

2
  • In this case it sounds like they have two distinct sets of network shares (home directories, mount points) in two (or more?) separate environments
    – voretaq7
    Apr 11, 2014 at 22:06
  • @voretaq7 Kinda seems to me like he's focused on or only cares about the network share aspect, not the different server groups... so, if all you're talking about is the shared folders/directories, use the terminology for that aspect, and ignore that the servers are in different logical groupings. (Say, what's wrong with calling these different groups of servers "groups of servers?") Apr 12, 2014 at 1:14
1

Given the fact we are referring to a specific site with a set of servers cooperating to perform a task the word 'farm' comes to mind. However this is a fairly specific term and requires a: the servers be on one site and b: the servers are cooperating in some way to perform a task.

1

What about BANK of servers? Perhaps they're on the same subnet, or maybe not, but you don't want them grouped by subnet in documentation.

A quick google search of the literal term shows it's not just me who thinks that way.

The literal term 'bank' when used as a noun means (according to dictionary.com):

a long pile or heap; mass: a bank of earth; a bank of clouds.

Vality has a point, 'farm' is also commonly used, as I almost always hear 'server farm' at work, and not 'server bank'. However, I hear 'bank of servers' more often than 'farm of servers', so it all depends on which one you want, just don't say 'server cloud', at all.

2
  • 1
    "Farm" is often used to mean something else. Often a loose cluster for redundancy/LB, like Citrix farm or web farm - or aggregation of resources, like a render farm.
    – mfinni
    Apr 12, 2014 at 18:42
  • @mfinni That's true, Citrix farm is common use, farm often has a double meaning. I suppose OP could just use the less fancy term Group instead. I don't hear Citrix farm much at work b/c we don't use Citrix!
    – MDMoore313
    Apr 12, 2014 at 19:11
0

Cluster of servers that serve specific shares, vs an additional cluster of servers that serve distinctly different shares.

8
  • I respectfully disagree : "Cluster" means a few specific things, which this usage doesn't really fit.
    – mfinni
    Apr 11, 2014 at 19:48
  • 1
    I'm not sure how you can disagree, as the relevance is the context in which the document-er is trying to convey his ideas. It's not your place to disagree, and in any event the definition of cluster is a number of things of the same kind. So a group of servers serving the same network shares, or same group of network shares is a cluster by definition.
    – kalikid021
    Apr 11, 2014 at 20:00
  • 1
    Loadbalancing is not a function of clustering, however in some types of clustering it may involve loadbalancing. The same for failover, etc. It's confusing because you are making assumptions on the word cluster based on what you are, or may be doing in your work or life with clustering. The poster is making documentation, which is explaining the details of his infrastructure. The documentation will include details of what he is describing. Using the word cluster infers a grouping of some sort, and not a specific set of technologies that somehow means cluster.
    – kalikid021
    Apr 11, 2014 at 20:20
  • 4
    For documentation purposes, it is good to stick to established norms for terminology, rather then inventing your own context or meaning. This helps maintain a common language among people who do the same work. mfinni is right that 'cluster' implies load-balancing/HA/failover capability.
    – jlehtinen
    Apr 11, 2014 at 20:43
  • 1
    @kalikid021, you are kind of making up your own terminology here. I'll admit this isn't definitive or exhaustive, but in my experience, Wikipedia summarizes it nicely. en.wikipedia.org/wiki/Computer_cluster#Attributes_of_clusters - load-balancing, compute (grid or similar), or HA/FT are the most common types of clusters. If you use the word to mean something else, you will confuse most other folks in the general IT field.
    – mfinni
    Apr 11, 2014 at 20:50

Not the answer you're looking for? Browse other questions tagged .