Overview
He best practices , then there reality ... they told me that for years. Because I think too many people think some best practices are set in stone and should be implemented or followed, regardless of the situation. To be honest, this could not be further from the truth. That's why many professional services organizations (CCS included) use the phrase "it depends." There is also a reason why our legal team at Citrix cringes when we use the term "best practice" ... because to a certain location in a certain client with a certain set of business requirements, this practice we recommended might not really be "better" or really optimal, and it could result mean downtime, disaster, unhappy customers, lost revenue, etc. and the word "best" may end up having legal implications, which is why we tend to use "best practices" in our documentation instead these days. But the point is better or best practices change ... some best practices change over the years ... a "best practice" in a situation with a client could be a bad practice to another customer. It really depends.
For this article, we begin 2014 and as I was doing some thinking on my almost 10 years at Citrix, I thought about what particular Citrix best practices have changed the most over the years. What best practices are old flat-out wrong these days? None of them are actually "new" as the title of my article might imply - but I hear a lot of these "myths" or old common best practices being said or implemented every day. So what follows is a collection of some of the most popular myths and best practices that have most changed dramatically in recent years in this world we live Citrix. I hope this finds everyone well into the new year and please do not be afraid to challenge the leading practices in the future ... the best practices are for change and I create new practices every day. 😉 (The list below is in no particular order by the way.)
Common Myths and "New" Best Practices
- session Reliability is Bad. I can not tell you how many times in the past, I recommend disabling PMC or the "Session Reliability". Honestly used to encode poorly and eventually causes excessive network traffic and hiding real problems network, while providing little or no use to the end user comments. But after our smart engineers in the UK got a hold of it a few years ago we moved to a less-IMA architecture XD in v5, CGP has changed. and it changed in the right direction. Now we recommend leaving SR / PMC enabled! Check this and this for more info.
- XenApp VM 2 vCPUs. this was another practice that has for some time there are about 5 or 6 years. But with recent advances in hardware schedulers hypervisor and NUMA awareness, I think we have finally proven it wrong. We are recommending XA VM with larger features VM all the time these days, such as 3, 4 or even 8 vCPUs. Check this and this for more info.
- PVS Must be physical. Our original position on PVS was to physics. But after setting 10Gb + network, progress around things like LACP and understand how PVS so we can properly size, we almost always recommend virtualizing PVS these days. Check this and this for more info.
- Isolate the PVS traffic flow. This is still somewhat controversial, especially because we dated technotes saying you should do this for performance or troubleshooting reasons (both invalid reasons in my opinion by the way). But again, with recent advances in virtualization and networking, I would say that there is little to gain from it. Our customers who keep their simple designs usually have the most success with PVS. Find out for more.
- Only Redirect Some Shell Folders . Not really specific Citrix, and more specific MSFT, but always something I wanted to address from the profile design is near and dear to my heart. We used to recommend that the lever folder redirection for specific records, such as shell and MyDocs AppData (and perhaps in the office, in some cases). But since MSFT re-designed profile namespace there are about 5 years and even told everyone they recommend to redirect all shell folders, we said the same thing. You can discuss the merits of the reorientation AppData, but is a bit beside the point here - we want you redirecting nearly everything you can so that 'ntuser.dat is roaming in and out
- MCS! You can not scale. I can still hear it said by many people and they do not even know why half the time. MCS can evolve. Especially if you pair it with thin provisioning and IntelliCache. And it does not require or generate 1.6x IOPS compared to MCS as we thought a few years ago (it is more like today 1.2x). Yes, there are still some operational challenges associated with MCS use in a part of the company ... but it absolutely can "scale" and it works fine. And it will get much better in the future, without the complexity of additional infrastructure and configuration introduced in PVS network.
- Multiple Farms are Bad. I tried to debunk this myth for years. XenApp several batteries are not a bad thing in my mind. Neither are many XD "sites" (really, they are firm, too - we just call them sites). I'm a big believer in horizontal scalability and "pod" architecture for scalability and stability known. We have many tools to manage multiple farms XA these days, too. And with virtualization and hypervisors be mainstream, I will always argue that spinning and management of some additional controllers or controllers These worth over failure caused by trying to vertical scale. Find out for more.
- 20 virtual machines per LUN must be strictly observed . First, it is a good rule of thumb. But it applies block-based storage only (FC, iSCSI, FCoE, and so on). I see too many people still quote this number or the design using this rule when you use NFS, which is file-based. I also see people designing this rule in mind, but they use vSphere VAAI and a network capable of! Save yourself the nightmare of managing the management of hundreds of tiny LUN - you can probably go a little bigger. Check this and this for more info.
- Pagefile RAM Should be 1.5x. Thanks again to the assistant, MarkR to debunk this myth for us all. This best practice is probably a decade now, and should never, never be followed! Please ask if you even need the ability of a complete memory dump ... and then do some simple tests and understand what size to make the swap file in your particular environment. Find out for more.
- SSD and shared storage are the only answer. I still get a ton of questions about whether SSDs are good, bad or ugly ... and if shared or local storage is the way to go. The bottom line is SSDs are still expensive and they are not all created equal in terms of write performance and longevity (yes, prices have fallen and we made some improvements in recent years, but still ...). And shared storage arrays major iron suppliers will always be expensive. This is why many 3rd party companies have sprung up over the last 5 years, the likes of Whiptail, Nutanix, Nimble, GreenBytes, etc. This is also why I think the dynamic storage tiering with QoS and storage virtualization are the future ... many customers get smart in that area and go with a hybrid approach already in terms of combination of SSDs and spinning disks ... and perhaps even using local storage for VDI deployments altogether to save a ton of money (especially when HA or reliability of non-persistent desktops are little interest!). So before you simply buy SSD or the next table million of our friends there at EMC, ask yourself if you really need for your virtualization project. Take a look at the I / O workloads generate first. Analyze your needs again. Look at how it could affect your business model. Or can you perhaps take a more intelligent approach is different this time, contrary to what you have done for the last decade?
Wrap-Up
this last point is probably a good ending because it's sort of the point of this article - i really want to encourage everyone to challenge old ways of thinking and to question some of the best age-old practices. Sometimes we are wrong to Citrix. Sometimes our people CCS could follow or implement the best ancient practices. And if we do, please let us know about it and we will repair. Even leave me a comment below and I will make it my personal mission to repair. And then we'll get communicated throughout our organization and community that the world is a better place.
Have you another "myth" or best practices that you think has changed considerably over the years? If so, please leave me a comment below and I'll either answer or perhaps even add to the above list with an update of the article. I'm a big believer that this kind of transparency can help customers design and partners and implement better solutions Citrix compatible in the future. 2014 will be a great year.
Cheers, Nick
Nick Rintalan, Lead Architect, Americas Consulting, Citrix Consulting Services ( "CCS")
0 Komentar