{"id":448,"date":"2018-08-30T22:55:23","date_gmt":"2018-08-30T19:55:23","guid":{"rendered":"https:\/\/blog.arcak.net\/?p=448"},"modified":"2018-09-03T08:23:56","modified_gmt":"2018-09-03T05:23:56","slug":"vmware-vsan-depolama-politikasi-bazli-yonetim-spbm","status":"publish","type":"post","link":"https:\/\/blog.arcak.net\/index.php\/2018\/08\/30\/vmware-vsan-depolama-politikasi-bazli-yonetim-spbm\/","title":{"rendered":"VMware vSAN Depolama Politikas\u0131 Bazl\u0131 Y\u00f6netim (SPBM)"},"content":{"rendered":"

\u00d6nceki yaz\u0131mda<\/strong><\/a>\u00a0k\u0131saca ge\u00e7ti\u011fim Storage Policy Based Management (SPBM)\u00a0<\/strong>k\u0131sm\u0131n\u0131 detayland\u0131rmaya \u00e7al\u0131\u015faca\u011f\u0131m. SPBM<\/strong>‘de belirtilen koruma tipleri Datastore Default, per-VM ve per-VMDK<\/strong> olarak uygulanabilmektedir.<\/p>\n

Depolama Politikas\u0131 Kurallar\u0131<\/strong>\u00a0\u00f6zg\u00fcn kaynak i\u00e7in buray\u0131<\/strong><\/a> inceleyebilirsiniz.<\/p>\n

\"\"<\/p>\n

<\/p>\n

Primary level of failures to tolerate (PFTT) :\u00a0<\/span><\/strong>Sanal makine nesnesinin tolare edebilece\u011fi sunucu ve ayg\u0131t hatalar\u0131 say\u0131s\u0131n\u0131 tan\u0131mlar. n<\/strong> adet ar\u0131zaya tolerans gerekliyse her bir veri RAID5<\/strong> ve RAID6<\/strong> parite verisi dahil olmak \u00fczere n+1<\/strong> yere yaz\u0131l\u0131r. Sanal sunucu olu\u015fturulurken depolama politikas\u0131<\/strong> se\u00e7ilmezse, varsay\u0131lan politika se\u00e7ilerek devam edilir. Fault Domain<\/strong> yap\u0131s\u0131 mevcutsa 2n+1<\/strong> kadar kapasiteye ihtiya\u00e7 duyulur. \u00d6rne\u011fin 2 Fault Domain varsa 2(2)+1=5<\/strong>\u00a0birim kapasiteye ihtiya\u00e7 duyulur. <\/span><\/p>\n

Stretched Cluster<\/strong>\u00a0(SC)<\/strong> yap\u0131s\u0131nda bu de\u011fer\u00a0 tolare edilebilecek sanal sunucu nesnelerini tan\u0131mlar. Her iki lokasyon aras\u0131nda veri korumas\u0131n\u0131 sa\u011flayabilmek i\u00e7in PFTT<\/strong> ve SFTT<\/strong> e\u015f zamanl\u0131 olarak kullan\u0131l\u0131r. SC\u00a0<\/strong>yap\u0131s\u0131nda\u00a0PFTT\u00a0<\/strong>maksimum de\u011feri 1<\/strong>‘dir.<\/strong>
\n<\/span><\/p>\n

Secondary level of failures to tolerate (SFTT) :\u00a0<\/strong>Stretched Cluster<\/strong> yap\u0131s\u0131nda bu kural bir lokasyondaki PFTT<\/strong> say\u0131s\u0131na ula\u015ft\u0131ktan sonra (lokasyon \u00e7\u00f6kt\u00fckten sonra)<\/strong> ikinci lokasyonda tolare edilebilecek sunucu say\u0131s\u0131n\u0131 belirler. \u00d6rne\u011fin PFTT=1<\/strong> ve SFTT=2<\/strong> ve birinci lokasyona eri\u015fim durduysa, cluster ikinci lokasyonda 2<\/strong> sunucu kayb\u0131na kadar tolerans g\u00f6sterebilir.<\/span><\/span><\/p>\n

Data Locality :\u00a0Stretched Cluster\u00a0<\/span><\/strong>yap\u0131s\u0131nda PFTT=0\u00a0<\/strong>olarak belirlendi\u011finde \u00e7al\u0131\u015f\u0131r. None, Preferred<\/strong> ya da Secondary<\/strong> se\u00e7enekleri bulunur. K\u0131saca veri gecikmesi istenmeyen sanal sunucular\u0131n \u00e7al\u0131\u015fmas\u0131 belirli lokasyonlara s\u0131n\u0131rlanabilir. Varsay\u0131lan de\u011feri\u00a0None<\/strong>‘d\u0131r.<\/span><\/span><\/p>\n

Failure tolerance method :<\/strong> Veri replikasyonu methodunun Performans<\/strong> ya da Kapasite<\/strong> odakl\u0131 olup olmayaca\u011f\u0131n\u0131 belirler.\u00a0RAID1\u00a0<\/strong>performans odakl\u0131d\u0131r ve daha \u00e7ok kapasiteye ihtiya\u00e7 duyar, 4 ve \u00fczeri say\u0131da\u00a0<\/strong>sunucu mevcut\u00a0RAID5\/6 (Erasure Coding)\u00a0<\/strong>ve PFTT=1\u00a0<\/strong>olarak se\u00e7ilirse kapasite odakl\u0131 olan\u00a0RAID5\u00a0<\/strong>elde edilir. 6\u00a0ve \u00fczeri say\u0131da\u00a0<\/strong>sunucu mevcut\u00a0RAID5\/6 (Erasure Coding) <\/strong>ve\u00a0PFTT=2\u00a0<\/strong>olarak se\u00e7ilirse ek g\u00fcvenlik ve kapasite odakl\u0131\u00a0RAID6\u00a0<\/strong>elde edilir.<\/span><\/span><\/p>\n

Stretched Cluster<\/strong>\u00a0(SC)<\/strong> yap\u0131s\u0131nda SFTT<\/strong> yap\u0131land\u0131r\u0131ld\u0131\u011f\u0131nda bu de\u011fer RAID<\/strong> yap\u0131s\u0131na de\u011fil tolare edilebilecek sunucu say\u0131s\u0131na etki eder.<\/span><\/p>\n

Number of disk stripes per object :<\/strong>\u00a0Her bir sanal makine nesnesinin ka\u00e7 ayg\u0131t \u00fczerine da\u011f\u0131t\u0131k olarak yaz\u0131laca\u011f\u0131n\u0131 belirler. Varsay\u0131lan 1<\/strong>‘dir, bundan b\u00fcy\u00fck de\u011ferler daha iyi performans verir ancak daha fazla sistem kayna\u011f\u0131 kullan\u0131r. Varsay\u0131lan de\u011ferin de\u011fi\u015ftirilmesi \u00f6nerilmez.\u00a0<\/strong><\/span><\/span><\/p>\n

Flash read cache reservation :\u00a0<\/strong>Varsay\u0131lan olarak cache reservation<\/strong> e\u015fit olarak b\u00fct\u00fcn nesneler aras\u0131nda payla\u015ft\u0131r\u0131l\u0131r ve de\u011ferin de\u011fi\u015ftirilmesi gerekmez. Sorun \u00e7\u00f6zme amac\u0131yla kullan\u0131labilir. Varsay\u0131lan de\u011feri %0<\/strong> ve Maksimum de\u011feri %100<\/strong>‘d\u00fcr.<\/span><\/span><\/p>\n

Force provisioning :\u00a0<\/strong>Bu kural aktif oldu\u011funda PFTT<\/strong>, Number of disk stripes per object<\/strong> ve Flash read cache reservation<\/strong> politikalar\u0131 gereksinimleri kar\u015f\u0131lanmasa dahi yeni sanal makine olu\u015fturulabilir. Sa\u011fl\u0131kl\u0131 \u00e7al\u0131\u015fan bir vSAN ortam\u0131nda bu kural aktif edilmemeli yaln\u0131zca felaket gibi aciliyet i\u00e7eren durumlarda yeni kurulum yap\u0131labilmesi i\u00e7in kullan\u0131lmal\u0131d\u0131r.<\/span><\/span><\/p>\n

Object space reservation :\u00a0<\/strong>Sanal sunucu disk (vmdk)<\/strong> nesnelerinin kaplayaca\u011f\u0131 alan\u0131 belirler. Varsay\u0131lan de\u011fer %0 (thin)\u00a0<\/strong>ve maksimum de\u011fer %100’d\u00fcr (thick)<\/strong><\/span><\/span><\/p>\n

Disable object checksum :<\/strong>\u00a0Varsay\u0131lan olarak bu de\u011fer No<\/strong> olarak tan\u0131ml\u0131d\u0131r. vSAN<\/strong> u\u00e7tan uca veri b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fc kontrol eden, bozulma oldu\u011funda otomatik olarak onaran mekanizmalara sahiptir.\u00a0Bu de\u011fer Yes<\/strong> olarak tan\u0131mlan\u0131rsa, veri b\u00fct\u00fcnl\u00fc\u011f\u00fc do\u011frulanamaz.<\/span><\/span><\/p>\n

IOPS limit for object :\u00a0<\/strong>Bir nesneye tan\u0131ml\u0131 (\u00f6r. VMDK)<\/strong>\u00a0IOPS<\/strong> limitini belirler. IOPS<\/strong> hesaplan\u0131rken okuma ve yazma i\u015flemleri e\u015fit kabul edilir ancak cache hit oran\u0131<\/strong> ve ard\u0131\u015f\u0131kl\u0131k<\/strong> g\u00f6z \u00f6n\u00fcnde bulundurulmaz. E\u011fer bir diskin\u00a0IOPS<\/b>\u00a0limiti ge\u00e7ilirse, I\/O<\/strong> i\u015flemleri s\u0131n\u0131rland\u0131r\u0131l\u0131r. IOPS<\/strong> limit for object = 0<\/strong> ise limitler zorunlu de\u011fildir. vSAN, operasyonun ilk saniyesi ya da bir s\u00fcre i\u015flem yap\u0131lmad\u0131\u011f\u0131nda nesnenin IOPS limit de\u011ferinin iki kat\u0131na \u00e7\u0131kmas\u0131na izin vermektedir.<\/span><\/span><\/p>\n